Highlights
-
•
First large study evaluating PCSK9 inhibitors and retinal vascular occlusion risk.
-
•
PCSK9 inhibitor therapy was associated with significantly lower RVO and RAO incidence.
-
•
Protective association persisted across 3-, 5-, and 7-year follow-up periods.
-
•
Results suggest a potential microvascular benefit of PCSK9 inhibition.
Purpose
PCSK9 inhibitors (PCSK9i) are a newer class of lipid-lowering drug that may be effective at lowering risk for retinal artery occlusion (RAO) and retinal vein occlusion (RVO). This study aims to investigate the relationship between PCSK9i use and retinal vascular occlusion among patients with hyperlipidemia.
Design
Retrospective, comparative clinical cohort study
Subjects, Participants, and/or Controls
Patients with hyperlipidemia, defined as serum low-density lipoprotein level of ≥130 mg/dL and total cholesterol level of ≥220 mg/dL, prescribed a lipid-lowering medication were identified. Patients prescribed a PCSK9i were included in the study group and compared with control patients prescribed any other type of lipid-lowering drug.
Methods
This study was conducted using electronic health record data from health organizations in the United States through the TrinetX platform. Propensity score matching was completed based on relevant patient demographics, comorbidities, and laboratory values. Comparison of main outcomes between the PCSK9i and non-PCSK9i groups was performed using measures of association analysis to determine risk ratio (RR) with 95% CI.
Main Outcome Measures
The outcomes measured consisted of occurrence of retinal vascular occlusion, RAO, RVO, central RAO, and central RVO at 3-year, 5-year, and 7-year time points.
Results
After propensity score matching, a total of 12,960 patients were included in each cohort. The analysis revealed that the PCSK9i cohort had a significantly lower risk for development of retinal vascular occlusions at multiple points, including 3-year (RR = 0.56, CI = 0.39-0.79), 5-year (RR = 0.50, CI = 0.37-0.67), and 7-year (RR = 0.46, CI 0.35-0.61) time points. This lower risk was also found in the PCSK9i group for an outcome of RVO at 5 years (RR = 0.50, CI = 0.34-0.73) and 7 years (RR = 0.47, CI = 0.33-0.67). For occurrence of RAOs (RR = 0.47, CI = 0.30-0.76) and central RVO (RR = 0.46, CI = 0.29-0.74) separately, the PCSK9i cohort had a lower risk at 7 years.
Conclusion
These findings suggest that PCSK9i may reduce the risk of retinal vascular occlusion compared with other classes of lipid-lowering medications.
INTRODUCTION
R etinal artery occlusions (RAO) are most often caused by an embolus composed of cholesterol, platelet fibrin, or calcium, or an in situ thrombosis. Meanwhile, retinal vein occlusions (RVO) are a result of thrombosis in the setting of venous stasis, hypercoagulability, and endothelial damage, as well as venous compression by the retinal artery. Loss of vision in patients with retinal vascular occlusions is the result of retinal ischemia and subsequent macular edema, placing patients at high risk for retinal detachment and neovascular glaucoma. Risk factors for retinal vascular occlusions include hypertension, carotid artery disease, cigarette smoking, dyslipidemia, obesity, and diabetes. Higher serum lipid levels are associated with altered plasma viscosity and platelet dysfunction, which may promote the development of retinal vascular occlusions.
Despite dyslipidemia being a significant risk factor for retinal vascular occlusions, there is no clear consensus on the effect of lipid-lowering drugs on the risk of RAO and RVO. Matei and associates found that statins had no preventive or therapeutic benefit on visual outcomes in high-risk patients with RVO after 43 months of follow-up. However, another study found that patients with RVO prescribed statins had a lower risk of future cardiovascular events, including stroke and myocardial infarction. Furthermore, 1 study evaluated the role of oral niacin in improving effects of chronic central RVO (CRVO), finding that patients on niacin experienced a reduction in cystoid macular edema and improved visual acuity after 1 year of therapy.
Although the effect of lipid-lowering medications on retinal vascular occlusions is not well established, it is known that they substantially reduce risk for atherosclerotic cardiovascular disease (ASCVD). The standard therapy, statins or 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR) inhibitors, mediates lipid levels by low-density lipoprotein cholesterol (LDL-C)– and non–LDL-C–dependent mechanisms. A new class of medications, PCSK9 inhibitors (PCSK9i), operates via a different mechanism for the treatment of hyperlipidemia in patients who are statin intolerant or require further reduction of cholesterol levels. PCSK9i decrease lipoprotein A levels and increase LDLR density on the cell surface of hepatocytes, thereby increasing uptake and metabolism of LDL-C. Higher levels of lipoprotein A are associated with the development of atherosclerosis and increased risk of stroke. Two meta-analyses found that PCSK9i decrease risk of ischemic stroke when compared with standard of care or placebo in patients at high risk of ASCVD, , suggesting that these medications may be even more effective than the standard therapy at reducing the risk of other emboli, such as RAOs.
Although the current literature demonstrates some evidence on the therapeutic benefit of statins for the prevention of RVO, there is a lack of research on this newer class of medications, PCSK9i, and its potential benefits for prevention of retinal disease. The purpose of this study is to evaluate the relationship between PCSK9i use and development of retinal vascular occlusions in patients with hypercholesterolemia.
METHODS
DATA SOURCE
This study used the United States (US) Collaborative Network within the TriNetX platform, using real-time data from the electronic health records (EHRs) of >117 million patients and 68 health care organizations across the US. This platform includes information on diagnoses, treatments, medications, laboratory tests, and genomic profiles. The data were aggregated, deidentified, and retrieved using Unified Medical Language System ontologies such as International Classification of Diseases (ICD), Ninth and Tenth revision code, current procedural terminology (CPT), and RxNorm. This retrospective study is exempt from informed consent. The data reviewed are a secondary analysis of existing data, do not involve intervention or interaction with human subjects, and are deidentified per the deidentification standard defined in Section 164.514(a) of the HIPAA Privacy Rule. The process by which the data are deidentified is attested through a formal determination by a qualified expert as defined in Section 164.514(b)(1) of the HIPAA Privacy Rule. This formal determination by a qualified expert refreshed on December 2020. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
PATIENT SELECTION
This study included patients aged >40 years based on likelihood to develop a retinal vascular occlusion. Initial cohorts were identified based on presence of hyperlipidemia or hypercholesterolemia, defined by patients having a serum LDL level of at least 130 mg/dL and a serum total cholesterol level of at least 220 mg/dL, as these patients are considered to be high risk. A serum LDL level of 130 mg/dL was chosen as patients with levels greater than this value are considered clinically to have hypercholesterolemia. Likewise, a serum total cholesterol level of 220 mg/dL was determined to be the cutoff as it is a halfway between the clinical thresholds of borderline high and high cholesterol. To ensure that patients had access to ophthalmology services, patients in both experimental and control groups were required to have an ophthalmology visit (CPT 1012793), encounter for examination of vision (ICD Z01.0), and diagnosis of an intraocular lens (ICD Z96.1) or an age-related cataract (ICD H25). Patients prescribed either PCSK9i, Evolocumab (Rx 1665684) or Alirocumab (Rx 1659152), were included in the “PCSK9i” cohort. Patients not prescribed a PCSK9i but any other lipid-lowering medications, including statins, niacin, fibrates, bile acid sequestrants, ezetimibe, and omega-3 acids, were included in the “non-PCSK9i” cohort. All medications are listed in Supplementary Table 1. The index event was defined as the first instance patients met both inclusion criteria, consisting of a patient’s first prescription of either PCSK9i drug or other lipid-lowering drug and laboratory values of serum LDL at least 130 mg/dL and total cholesterol at least 220 mg/dL. The CONSORT diagram demonstrates the number of patients at each step of the inclusion and exclusion processes (Supplementary Figure 1). Patients with a history of RVO and/or RAO, based on ICD encounter diagnoses, were excluded from this study. The TrinetX US Collaborative Network has data considered reliable going back 20 years.
Propensity score matching (PSM) was completed to reduce confounding variables and bias and was based on demographics, including sex, race, ethnicity, and age at index event, familial hypercholesterolemia, ASCVD risk, most recent patient body mass index, and risk factors for RVO. All ICD-10-CM, CPT, and RxNorm codes are listed in Supplementary Table 1. Matching was also performed on the patients’ most recently recorded laboratory values, including serum LDL cholesterol, HDL cholesterol, total cholesterol, and total triglycerides. Baseline diagnoses and patient characteristics were collected 1 year before index event. Irritant contact dermatitis (ICD-10 L23) was used as a negative control for each analysis to ensure the legitimacy of these comparisons, as this outcome is not likely to be affected by confounding factors. PSM was completed for each analysis at all time points ( Table 1 , Supplementary Tables 2-10).
TABLE 1
Demographics and Comorbidities for Patients in the PCSK9i vs Non-PCSK9i Cohorts at 3 Years Before and After Propensity Score Matching.
| Variable: Avg. ± SD or n (%) | Before Propensity Score Matching | After Propensity Score Matching | ||||
|---|---|---|---|---|---|---|
|
PCSK9i
( n = 12,398) |
Non-PCSK9i
( n = 562,327) |
Standardized Mean Difference |
PCSK9i
( n = 12,383) |
Non-PCSK9i
( n = 12,383) |
Standardized Mean Difference | |
| Demographics | ||||||
| Age at index | 67.2 ± 9.4 | 63.7 ± 11.7 | 0.326 | 67.2 ± 9.4 | 67.2 ± 10.2 | 0.002 |
| Female | 7572 (61.1) | 330,740 (62.3) | 0.025 | 7560 (61.1) | 7584 (61.2) | 0.004 |
| Male | ||||||
| White | 8787 (70.9) | 348,539 (65.6) | 0.113 | 8772 (70.8) | 8872 (71.6) | 0.018 |
| Black | 1834 (14.8) | 90,866 (17.1) | 0.063 | 1834 (14.8) | 1821 (14.7) | 0.003 |
| Hispanic or Latino | 621 (5.0) | 39,886 (7.5) | 0.103 | 621 (5.0) | 511 (4.1) | 0.043 |
| Comorbidities | ||||||
| Primary hypertension | 9428 (76.0) | 311,878 (58.7) | 0.369 | 9413 (76.0) | 9363 (75.6) | 0.009 |
| Type 2 diabetes mellitus | 5478 (44.2) | 171,377 (32.3) | 0.247 | 5468 (44.2) | 5434 (43.9) | 0.006 |
| Tobacco use | 490 (4.0) | 13,165 (2.5) | 0.084 | 490 (4.0) | 435 (3.5) | 0.023 |
| Nicotine dependence | 1193 (9.6) | 45,953 (8.7) | 0.034 | 1193 (9.6) | 1041 (8.4) | 0.043 |
| Coronary artery disease | 7063 (54.5) | 62,815 (11.5) | 1.028 | 6755 (54.6) | 6803 (54.9) | 0.008 |
| Overweight and obesity | 3685 (29.7) | 98,602 (18.6) | 0.263 | 3676 (29.7) | 3570 (28.8) | 0.019 |
| Labs | ||||||
| LDL cholesterol in serum | Mean 137.7 ± 57.3; 10,174 (82.1) | 127.5 ± 43.2; 369,180 (69.5) | 0.202 | Mean 137.7 ± 57.3; 10,159 (82.0) | Mean 133.2 ± 50.3; 10,075 (81.4) | 0.082 |
| 0-115 mg/dL | 4472 (36.1) | 160,439 (30.2) | 0.125 | 4464 (36.0) | 4416 (35.7) | 0.008 |
| 115-130 mg/dL | 1809 (14.6) | 57,299 (10.8) | 0.114 | 1806 (14.6) | 1701 (13.7) | 0.024 |
| 130-145 mg/dL | 2039 (16.4) | 68,573 (12.9) | 0.100 | 2031 (16.4) | 1857(15.0) | 0.039 |
| 145-160 mg/dL | 1974 (15.9) | 64,763 (12.2) | 0.107 | 1962 (15.8) | 1800 (14.5) | 0.036 |
| 160-175 mg/dL | 1694 (13.7) | 48,835 (9.2) | 0.141 | 1685 (13.6) | 1588 (12.8) | 0.023 |
| >175 mg/dL | 3263 (26.3) | 60,251 (11.3) | 0.390 | 3248 (26.2) | 3345 (27.0) | 0.018 |
| HDL cholesterol in serum | Mean 47.5 ± 20.2; 10,236 (82.6) | 50.1 ± 20.2; 371,504 (70.0) | 0.129 | Mean 47.5 ± 20.2; 10,221 (82.5) | Mean 47.7 ± 20.6; 10,103 (81.6) | 0.007 |
| Total cholesterol in serum | Mean 222.6 ± 66.1; 10,079 (81.3) | Mean 210.6 ± 50.8; 369,865 (69.7) | 0.204 | Mean 222.6 ± 66.1; 10,066 (81.3) | Mean 214.9 ± 59.1; 9942 (80.3) | 0.122 |
| Total triglycerides in serum | Mean 179.9 ± 143.9; 10,206 (82.3) | Mean 158.5 ± 124.7; 364,059 (68.6) | 0.159 | Mean 179.8 ± 143.9; 10,191 (82.3) | Mean 161.3 ± 127.4; 10,085 (81.4) | 0.137 |
| Body mass index | Mean 30.5 ± 6.2; 9820 (79.2) | Mean 30.2 ± 6.8; 366,352 (69.0) | 0.049 | Mean 30.5 ± 6.2; 9806 (79.2) | Mean 30.8 ± 6.8; 9810 (79.2) | 0.033 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
