The Interplay Between Ophthalmic and Systemic Outcomes in Patients With Sickle Cell Disease and Concurrent Retinopathy—A Population-Based Study

Abstract

Sickle cell disease (SCD) is the most common hemoglobinopathy worldwide caused by the presence of hemoglobin S or sickle hemoglobin [1]. It is estimated to affect 100 000 people in the United States and more than 3 million people worldwide [2]. Chronic and recurrent sickling, hemolysis, and endovascular inflammation results in pain and end-organ damage [3]. Patients with SCD suffer from acute and chronic systemic complications that range from stroke, acute chest syndrome, priapism, recurrent pain crises, and splenic sequestration to end-organ damage to the kidneys, liver, heart, and retina [1]. Sickle cell retinopathy (SCR) affects both the anterior and posterior segments of the eye. SCR occurs in around 20% of people with homozygous HB SS disease [4].

Previous studies have examined the incidence and treatment of SCR in SCD, but these studies have been limited to smaller cohorts in particular geographic areas. There is a lack of contemporary, large-scale, adult population-based studies that examine the interplay of the various systemic complications of SCD and SCR. The purpose of this communication is to compare the rates of various systemic complications, systemic medication use, hospitalization, and mortality rates associated with SCD, in patients with and without concurrent SCR.

A retrospective cohort study was performed using the TriNetX network, an electronic health records (EHR) research network comprising multiple healthcare organizations across the USA and globally. The database gives access to demographic details, diagnosis, procedures and laboratory values over 20 years (2004–2024). We included patients greater than 18 years of age, with at least 5 years of follow-up. Patients were identified using the International Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10). Cohort 1, included patients with the diagnosis of SCD (H57.1) and SCR (H36 and H35.2) (Table S1). Patients were excluded if they had a diagnosis of sickle cell trait (SCT) (D57.3), pre-existing age-related macular degeneration (ARMD) (H35.32), or diabetes mellitus with ophthalmic complications (E10.3, E11.3). Cohort 2 included patients with the diagnosis of SCD without SCR. Patients were excluded if they had ICD codes for SCT, SCR, ARMD, and diabetes mellitus with ophthalmic complications (Table S2). Current procedural terminology (CPT) codes were used to identify the medical and surgical procedures performed. For patients in cohort 1, the diagnosis of SCR was considered the index event; for patients in cohort 2, the diagnosis of SCD was considered the index event.

Descriptive analysis was performed between cohorts 1 and 2 to compare rates of systemic complications, hospitalization rates, mortality rates, and systemic medication use at 1 and 5-year follow-up (Table S3). Propensity score matching (PSM) was performed between the two cohorts to control for age, sex, race, ethnicity, hypertension, hyperlipidemia, and nicotine use, which resulted in 2 comparison cohorts. PSM was performed using the TriNetX built-in analysis platform (1:1 matching by nearest neighbor greedy matching algorithm with a caliper of 0.25 SDs).

Baseline demographics of both the cohorts before and after PSM are outlined in Table S4. Each cohort had 1249 patients, the mean age at index event was 34.9 ± 14.8 years and 35.2 ± 15.1 years in cohorts 1 and 2 respectively, with females comprising 51.4% and 52.6% of the population in cohorts 1 and 2 respectively. The majority identified as Black or African American (87%) in both cohorts. The average follow-up period in both cohorts was 8 years.

We compared the rates of systemic complications in cohorts 1 and 2 (Table 1). The most common complication was sickle cell crisis; the rates were two-fold higher in cohort 1 as compared to cohort 2 at 1 year (54.1% vs. 29%; risk ratio (RR), 1.87; p < 0.01) and 5 years (73.6% vs. 43.2%; RR, 1.70; p < 0.01). The risk of avascular necrosis (AVN) was two-fold higher in cohort 1 as compared to cohort 2 at 1 year (17.5% vs. 5.1%; RR, 3.4; p < 0.01) and 5 years (32.8% vs. 13.7%; RR, 2.3; p < 0.01). The risk of acute chest syndrome (ACS) was almost three-fold higher in cohort 1 as compared to cohort 2 at 1 year (8.6% vs. 3%; RR, 2.87; p < 0.01) and 5 years (24.5% vs. 10%; RR, 2.48; p < 0.01). The risk of myocardial infarction (MI) was similar in both cohorts, at 1 year (0.8% vs. 0.9%; RR, 0.91; p = 0.83) and 5 years (3.1% vs. 3.2%; RR, 0.95; p = 0.82). The risk of stroke was similar in both cohorts at 1 year (4.2% vs. 4%; RR, 1.08; p = 0.69), but was higher in cohort 1 at 5 years (8.7% vs. 7%; RR 1.24; p = 0.12).

Risk of sickle cell nephropathy was two-fold higher in cohort 1 at 1 year (3.2% vs. 1.7%; RR, 1.86; p = 0.02) and 5 years (8% vs. 5.1%; RR, 1.56; p < 0.01). The risk of opioid dependence was two-fold higher in cohort 1 at 1 year (4.1% vs. 1.8%; RR 2.26; p = 0.001) and 5 years (10.1% vs. 5.6%; RR, 1.76; p < 0.01). The risk of sickle cell-related priapism was similar in both cohorts at 1 year (3.2% vs. 1.7%; RR, 1.90; p < 0.01) but the risk was higher in cohort 1 at 5 years (6.4% vs. 3.7%; RR, 1.70; p < 0.01). The risk of hospitalization was almost two-fold higher in cohort 1 at 1 year (29.3% vs. 17.6%; RR, 1.66; p < 0.01) and 5 years (50% vs. 33.3%; RR, 1.53; p < 0.01). The risk of mortality was similar in both cohorts, at 1 year (1% vs. 0.8%; RR, 1.30; p = 0.53) and 5 years (3.3% vs. 2.2%; RR, 1.50; p = 0.09).

We also analyzed the rates of systemic medication use in both cohorts. The most commonly prescribed drug was hydroxyurea, with a two-fold higher rate in cohort 1 as compared to cohort 2 at 1 year (29% vs. 16%; RR, 1.80; p < 0.01) and 5 years (41.3% vs. 27%; RR, 1.54; p < 0.01). The rates of other systemic medications L-glutamine, crizanlizumab, and voxelotor were also higher in cohort 1. (Table S5).

We present a comprehensive analysis of systemic complications, systemic medication use, hospitalization, and mortality rates in patients with SCD, with and without SCR. Sickle cell crisis, also known as vaso-occlusive crisis (VOC), is the classical systemic complication of SCD and leads to high rates of hospitalization, morbidity, and mortality [5]. In our study, the most common systemic complication was sickle cell pain crisis, with a two-fold higher incidence in sickle cell patients with known SCR. A close relationship exists between VOC and ACS, the most common cause of death in patients with SCD [6]. We report a three-fold higher incidence of ACS in our patients in cohort 1 as compared to cohort 2, suggesting that SCD patients with SCR are at an elevated risk for ACS. These findings suggest that presence of SCR may indicate more widespread and severe vascular involvement and thus more concurrent systemic complications.

Our study also investigated hospitalization and mortality rates among patients with SCD, with and without SCR. Hospitalization rates were almost two-fold higher in cohort 1 as compared to cohort 2, highlighting an increased healthcare burden after SCR is diagnosed. In this study, the most commonly prescribed drug in both cohorts was hydroxyurea, followed by L-glutamine. All systemic medications were prescribed at higher rates in cohort 1 as compared to cohort 2, suggesting that SCD patients affected by retinopathy may have a greater need for systemic medical therapy for their SCD.

The rates of other systemic complications like sickle cell nephropathy, sickle cell-related priapism, opioid dependence and stroke were also significantly higher in patients with SCR as compared to those without SCR. While prior studies have focused on the ocular aspects of SCR, our findings underscore the impact of SCR diagnosis on the systemic health outcomes of SCD patients. These findings suggest that SCR may be a marker of the overall systemic burden of SCD, resulting in more frequent and severe systemic complications including the need for hospitalizations and systemic medications.

This study is subject to inherent limitations associated with the analysis of large, de-identified aggregated medical health records data. It relies on ICD-10 diagnosis and coding, which may introduce inaccuracies. Although the HB-SC genotype has more severe SCR [7], but fewer severe systemic complications, we excluded patients with HB-SC genotype, as our primary aim was to compare systemic complications in SCD patients with SCR and without SCR. In order to capture any possible retinopathy associated with SCR, we included patients with ICD-10 codes H35 (Other non-diabetic proliferative retinopathy) and H36 (Retinal disorders in diseases classified elsewhere) for SCR, which may cause a possible misclassification of non-SCR retinal diseases. Further, we did not stratify patients based on the severity of SCR. The database lacks laterality data and access to individual patient records and operative notes. However, our study benefits from a diverse and representative study population, enhancing the generalizability of our findings.

To our knowledge, this is the first and the largest study to compare systemic complications between SCD patients with and without SCR. For ophthalmologists and hematologists, it is crucial to recognize the interconnectedness between ocular findings and systemic complications in SCR patients. This work may help educate patients as well as to the importance of medication compliance and systemic follow-up particularly when SCR is identified. SCR may serve as a marker for more severe systemic disease, emphasizing the importance of multidisciplinary collaboration with hematologists and internists to ensure comprehensive patient care.

Authorship: All authors attest that they meet the current ICMJE criteria for Authorship.

The authors report no relevant financial disclosures or proprietary interests in the materials described in the article.

The study involved de-identified patient data from the TriNetX database and was exempt from institutional review board approval.

The study involved de-identified data from the TriNetX database and does not require patient consent.

The authors declare no conflicts of interest.