Risk of glaucoma after vitreoretinal surgery – Findings from a population-based cohort study

. Purpose: To investigate the association between diﬀerent types of vitrectomy and risk of diﬀerent types of glaucoma and to determine the eﬀect of systemic medication and diabetes status on this risk. Methods: A population-based nested case – control study included individuals of age ≥ 18 years who had undergone single vitrectomy, vitrectomy with retinal procedure, or combined phaco-vitrectomy between 2001 and 2010. End of follow-up was 2017. Odds ratio (OR) for the development of glaucoma after diﬀerent types of vitrectomy and 95% conﬁdence interval (CI) were based on conditional logistic regression models. For every glaucoma case, ﬁve controls were matched by age, sex, start of follow-up year, and hospital district. Results: The individuals diagnosed glaucoma and 31 135 controls matched age, sex, and hospital district. Vitrectomy was performed 103 eyes in the glaucoma group and 158 eyes in the control group. As regards the risk of any glaucoma, the risk was lowest in eyes underwent combined phaco-vitrectomy (OR: 2.7, 95% CI: 1.8 4.1), followed by single vitrectomy (OR: 3.15, 95% CI: 2.1 4.8), and highest in eyes that underwent vitrectomy with retinal procedure (OR: 4.5, 95% CI: 2.7 – 7.4). Diabetes had no eﬀect (OR: 0.96, 95% CI: 0.92 – 1.01), but 5-year systemic statin use slightly decreased glaucoma risk (OR: 0.86, 95% CI: 0.77 – 0.97). Conclusions: Vitreoretinal surgery was associated with an increased glaucoma risk; the risk being related to the complexity of vitrectomy. Long-term systemic statin therapy may decrease glaucoma risk, while diabetes had no association.


Introduction
Glaucoma is the leading cause of irreversible blindness globally. The risk and subtypes of glaucoma vary in different populations. Currently, the number of people aged 40-80 years with glaucoma worldwide is 64.3 million but expected to increase to more than 110 million by 2040 (Tham et al. 2014).
Primary open-angle glaucoma (POAG) is characterized by degeneration of retinal ganglion cells often associated with an increase in intraocular pressure (IOP) due to hindered aqueous humour drainage through the trabecular meshwork (TM) and uveoscleral pathway. Recent studies have highlighted the role of lowgrade inflammation, oxidative stress and damage to the TM in the development of OAG (Weinreb et al. 2014;Siegfried & Shui 2019). On the other hand, the aetiology of secondary glaucoma (SG) is more complex, including inflammation, lens-and drug-induced responses, retinal vascular diseases, certain syndromes, tumours, trauma, and/or anterior and posterior segment surgery. In recent years, the epidemiology of glaucoma subtypes has changed, showing increase in drug-induced, anterior segment surgery-related and syndrome-associated SG cases (Gong et al. 2021).
Vitreoretinal surgery has advanced extensively during the last decades, being currently the third most performed eye surgery after cataract and refractive surgery (El-Amir et al. 2009;Wubben et al. 2016). Some vitreoretinal surgery manoeuvres, such as stripping of the internal limiting membrane (ILM), can lead to adverse events including M€ uller cell dysfunction, eccentric paracentral macular holes, macular microscotomata, and retinal dimpling (D ıaz-Valverde & Wu 2018). The perioperative use of dyes, retinal photocoagulation, cryotherapy, intravitreal anti-VEGF, use of fluid-airexchange/perfluorocarbon/gas/silicone oil (SO) and other treatments may have an impact on long-term visual outcomes and development of glaucoma after vitreoretinal surgery.
Today, the precise pathogenesis of glaucoma after vitreoretinal surgery remains unknown. In addition, the treatment of glaucoma after vitreoretinal surgery may be challenging due to noncompliance to conventional therapies (Koreen et al. 2012). Furthermore, systemic diseases such as diabetes per se or systemic medication may play a role in the individual risk of postvitrectomy glaucoma (Feldman-Billard & Dupas 2021).
Generally, it has been reported that there is an increased risk of OAG after pars plana vitrectomy (Chang 2006;Mansukhani et al. 2018;Miele et al. 2018). However, findings related to association of glaucoma with vitreoretinal surgery have been heterogenous, suggesting that further studies need to be conducted in more homogenous cohorts (Miele et al. 2018). Currently, there is a strong need to investigate the association of less complication prone macular surgery (vitrectomy) and more complication prone complex vitreoretinal surgery (vitrectomy with retinal procedure) with the risk of glaucoma. Also, evaluating the risk of glaucoma after combined phaco/lensectomyvitrectomies is warranted.
The main objective of this longitudinal population-based study was to investigate the association between different types of vitreoretinal surgery and risk of glaucoma. We examined the risk of OAG or other type of glaucoma in eyes that underwent three types of vitreoretinal surgery based on the complexity of posterior segment pathology.

Methods
Study design was longitudinal historic population-based. The original FIN-CARING2 study population included 398 708 individuals. Of them, 199 354 subjects had diabetes, and an equivalent reference population was nondiabetic, and matched by age, sex and region, as described (Niskanen et al. 2020). Subjects with diabetes had insulin or oral antidiabetic prescriptions and medication reimbursement claims.

Inclusion criteria
Out of 244 100 individuals selected for the study of glaucoma incidence, we constructed a nested case-control study population aiming to sample five controls (n = 31 135) for each glaucoma case (n = 6552), matched by age (within 1 year), sex, start of follow-up calendar year (within 1 year), and hospital district. In the analysis, some cases had less than five controls. The start of follow-up was from January 1, 2001, to December 31, 2010, and the end of follow-up was December 31, 2017, covering a 17-year study period. Individuals with a start date before January 1st 2001 were excluded in order to secure a 5-year history of hospital or tertiary health care records for the whole study population before the start of follow-up. Individuals with already diagnosed glaucoma or who had undergone vitreoretinal surgery before start of follow-up were excluded.
The other glaucoma cases included unspecified glaucoma (H40.2-8) or glaucoma secondary to other eye disorders. First, we analysed individuals who were diagnosed with any glaucoma (a). Next we subgrouped glaucoma cases to (b) OAG or (c) other type of glaucoma taking into account the exact type of vitreoretinal eye surgery.
We aimed to identify the exact ICD-10 diagnoses of vitreoretinal surgeries and evaluate the risk of glaucoma within each of the three groups (i.e. CKD91, CKD92, and CKD94). Since 38 out of 261 surgeries (14.5%) missed accurate ICD-10 diagnosis, this additional data could not be analysed reliably.

Exclusion criteria
Individuals who underwent a combined procedure on vitreous body and retina including encircling scleral buckle/explant or band (CKD93) were excluded due to complex retinal pathology and vitreoretinal surgery. Encircling band per se has the potential to compress episcleral veins and cause IOP elevation.

Systemic medication and comorbidities
Data on medication prescriptions and reimbursements were obtained from the Social Insurance Institute, which provides national public health insurance to all residents in Finland (Kela 16/522/2012). We identified systemic medications at baseline by dispensed prescriptions and comorbidities by granted special medication reimbursements. Systemic medications included angiotensin agents, calcium channel antagonists, beta-blockers, diuretics, metformin, statins, selective serotonin reuptake-inhibitors (SSRI), tricyclic antidepressants and drugs for Alzheimer's disease. Information of the following systemic diseases were collected: thyroid insufficiency, adrenocortical insufficiency, severe psychotic and other severe mental disorders, cancer, chronic cardiac insufficiency, chronic hypertension, chronic coronary heart disease, and chronic arrhythmias. Altogether, the following baseline variables were available: age, sex, diabetes (yes/ no), socioeconomic group, other systemic comorbidities, systemic medication, and hospital district.

Main objective
The main objective was to study the risk of new onset OAG or other type of glaucoma after vitreoretinal surgery. Special interest was paid to the exact type of vitreoretinal surgery (CKD91, CKD92, CKD94), as well as diabetes status and exposure to statin therapy before and at the time of surgery.

Ethical considerations
We obtained the approval from the Ethics Committee of Faculty of Medicine, University of Helsinki, 17 January 2012 (Ref 02/2012). Approvals to extend the original study plan and research group were granted by register holders (the Social Insurance Institute (Kela 29/ 522/2019), and the Institution for Health and Welfare (THL/486/5.05.00/2019 and THL/3157/14.02.00/2020)).

Statistical analysis
We modelled the risk of glaucoma after vitreoretinal surgery using a conditional logistic regression model that takes matching into account. Odds ratio (OR) and 95% confidence interval (CI) are reported. All analyses were carried out separately for the two glaucoma endpoints (OAG and other glaucoma).
Operated eyes were divided into three groups according to the vitrectomy procedure received: Vit (CKD91, single vitrectomy), VitRet (CKD92, combined vitrectomy with retinal procedure), and PhacoVit (CKD94, combined phaco-vitrectomy). The analysis comparing the association of statin use was conducted using vitrectomy subgroups. Calculations were carried out with R language (R Core Team 2019).

Results
Of the study population (n = 37 687), 52.8% were female (n = 19 895) ( Table 1). Patients were operated in 21 districts in Finland. A quarter of the population was from the capital region, Helsinki and Uusimaa district (n = 9639; 25.6%). Out of all glaucoma cases, 33.8% (n = 2216) were diagnosed as OAG and 66.2% (n = 4336) were of other glaucoma types.
Based on the conditional logistic regression model, the risk of any glaucoma (OAG or other glaucoma type) was increased after all three types of vitreoretinal surgeries, but there were clear differences between the vitreoretinal surgery groups ( Table 2). The risk of any glaucoma was lowest in the eyes that underwent combined phaco-vitrectomy (PhacoVit; OR: 2.7, 95% CI: 1.8-4.1), followed by single vitrectomy (Vit; OR: 3.15, 95% CI: 2.1-4.8), being highest in the eyes that underwent vitrectomy with retinal procedure (VitRet; OR: 4.5, 95% CI: 2.7-7.4).
Diabetes was not associated with the risk of any type of glaucoma after vitreoretinal surgery (adjusted OR: 0.94, 95% CI: 0.87-1.03) and did not show any association in the glaucoma subgroup analyses either (not shown). When socioeconomic status was taken into account, the risk of glaucoma after vitreoretinal surgery was not increased in any socioeconomic group compared to upper-level employees (Table 2). Of note, the risk was lower in selfemployed, manual workers and pensioners.
Next, we included time since surgery as an additional predictor in the conditional logistic regression models as follows: no surgery, less than 1 year, 1-2 years, 2-5 years, over 5 years. However, we did not find any association between time since surgery and risk of glaucoma in our study (data not shown).

Exposure to statin therapy before surgery
Since statins have been suggested to reduce the risk of glaucoma (McCann et al. 2016), we found it reasonable to study whether lipid-lowering statin therapy modulates the risk profile of glaucoma after vitreoretinal surgery. Statin therapy was the third most common systemic medication within the cohort. Of the 7830 statin users, 6487 (20.8%) belonged to age-and sex-matched controls and 1343 were glaucoma cases (20.5%; p = 0.552). Cumulative exposure to statin therapy (0-5 years) before vitreoretinal surgery was analysed (Table 3). Five-year exposure to statin treatment slightly decreased the risk of any glaucoma (OR: 0.86, 95% CI: 0.77-0.97) (Table 4). However, exposure time shorter than 5 years was not associated with the risk of any glaucoma. According to the glaucoma subtype analysis, both 4-year exposure (OR: 0.87, 95% CI: 0.78-0.97) and 5-year exposure (OR: 0.81, 95% CI: 0.69-0.95) to statin therapy before vitreoretinal surgery slightly decreased the risk of other glaucoma, but no difference was found in the OAG cases (not shown). Shorter duration of statin therapy (less than 3 years) had no association with the risk of either OAG or other glaucoma after vitreoretinal surgery.

Main finding
In this nested case-control study, we analysed 261 individuals who had undergone one of three main types of vitreoretinal surgical procedures in Finland during years 2001-2010, and of whom 103 developed glaucoma and 158 were age, sex and hospital district matched controls. The risk of any glaucoma was increased after all three types of vitrectomy, confirming recent findings (Mansukhani et al. 2018;Miele et al. 2018). Of note, the risk of any glaucoma was highest among the eyes that underwent vitrectomy with retinal procedure (VitRet group) and lowest in the eyes that underwent combined cataract surgery and vitrectomy (PhacoVit group).

Results in context with the published literature
Vitreoretinal surgery has advanced extensively during the last decades, and combined cataract and vitrectomy procedures are commonly performed in Europe. The risk of any glaucoma being lowest in the group that underwent combined cataract and vitrectomy may be related to a tendency of IOP to decrease after cataract surgery (Mansberger et al. 2012;Masis Solano & Lin 2018). In addition, the eyes that undergo combined phaco-vitrectomy are often those with vitreoretinal interface diseases undergoing routine macular surgery. Previous studies have also shown that in eyes receiving macular surgery for disorders such as epiretinal fibrosis and/or macular hole, the condition is less inflammatory, but more neurodegenerative ( € Ohman et al. 2018). In our sub-analysis, patients undergoing combined phaco-vitrectomy had more than 4-fold risk of other glaucoma, but no risk association occurred among OAG eyes. This suggests that peeling of the membranes (epiretinal membrane and/or ILM) from the surface of the neural retina can damage the retinal ganglion cell layer, eventually leading to visual field defects and secondary glaucoma (Akino et al. 2020;Corvi et al. 2021). Moreover, combined phaco-vitrectomy prevents secondary narrowing of the anterior chamber angle due to thickening of crystalline lens post-vitrectomy, a complication often occurring in eyes that undergo only vitrectomy, making it understandable why risk of OAG was not increased.
On the other hand, in the eyes with more complex vitreoretinal pathology (such as DR or RRD), there is more inflammation at the time of surgery and after surgery Aptel et al. 2017;€ Ohman et al. 2021). Of note, glaucoma shares many similarities with both DR and RRD disease processes, making the pathology complex and interconnected (Karlstetter et al. 2015). In all these three disease processes, the mechanisms of injury include inflammation and abnormal protein accumulation (Karlstetter et al. 2015;Loukovaara et al. 2015;Williams et al. 2017;Dammak et al. 2021;€ Ohman et al. 2021).
Noteworthy, the complement cascade, the toll-like receptors, and tumour necrosis factor (TNF)-a pathway have been implicated in the pathogenesis of glaucoma, DR and RRD (Tezel 2008;Luo et al. 2010;Texel et al. 2010;Loukovaara et al. 2015;Dammak et al. 2021;€ Ohman et al. 2021). Taken together, the increased risk of glaucoma may be explained by oxidative stress, neuroinflammation, proteolytic degradation, degenerative changes, dysregulation of ocular hemodynamics, genetic factors, as well as aberrant cellular signalling known to be implicated in neurodegeneration and cell loss associated with both central nervous system and retinal disorders (Tezel 2008;Inyushin et al. 2019;Rodr ıguez Villanueva et al. 2020;€ Ohman et al. 2021).
Other less studied pathological mechanisms likely to be involved in the optic neuropathy or secondary glaucoma after vitreoretinal surgery could be related to newly discovered glymphatic system (Wostyn et al. 2015;Wostyn 2021). Post-vitrectomy, the rapid changes of IOP and composition of the vitreous substitute (whether gas or SO) as well as changes in cerebrospinal fluid surrounding the optic nerve could have impact on the pathogenesis of glaucoma. Decades ago, it was shown that fluids from the vitreous body and the optic nerve move from opposite directions and converge at the optic nerve head (Hayreh 1978). Thus, it is possible that vascular and glymphatic circulatory dysfunction may play a role in the pathogenesis of glaucomatous damage in vitrectomized eyes.
The more complex the vitreoretinal surgery, the more breakdown of inner and outer blood-retinal barrier there is in the posterior segment (Coca-Prados 2014). In addition, the more complex the posterior segment pathology, the more pro-inflammatory and pro-fibrotic cytokines and growth factors will be released both before surgery and during tissue manipulation at the time of surgery (Pastor et al. 2016;€ Ohman et al. 2021). As expected, in our sub-analysis, the risk of glaucoma was highest in the group that underwent vitrectomy combined with retinal procedure, the risk being 3.2-fold for OAG and 5.4-fold for other type of glaucoma. Herein, in addition to simple membrane peeling, intraoperative surgical manoeuvres and tissue manipulation such as use of perfluorocarbon, performed endolaser/ endocryo, performed retinotomy/ retinectomy, and type of endotamponade (whether gas or SO) will most likely increase the post-surgery risk of glaucoma.
In the eyes that have undergone complex vitreoretinal surgery, there are also other factors such as postoperative face-down positioning, which can cause IOP elevation and explain our findings compared to single vitrectomies. Of note, SO can block the pupil or migrate into the anterior chamber following vitreoretinal surgery. In eyes with long-term SO endotamponade, the emulsified SO droplets in the Schlemm's canal may predispose to IOP elevation and development of glaucoma post-vitrectomy (Choi et al. 2020;Nicolai et al. 2020;Aktas et al. 2021).
Diabetes was not associated with the risk of glaucoma after vitreoretinal surgery. Diabetic patients may have shallower anterior chambers and thicker crystalline lenses than nondiabetic patients (Kocat€ urk et al. 2014). Numerous studies have, however, shown that cataract surgery improves aqueous humour dynamics by lowering the IOP in both diabetic and non-diabetic eyes (Mansberger et al. 2012;Masis Solano & Lin 2018;Bayat & Akpolat 2021).
In this study, cumulative exposure to statin treatment longer than 4 years was associated with a slightly decreased risk of other type of glaucoma after vitreoretinal surgery, though no such association was found for OAG. This finding is of interest, since decreased risk of glaucoma could be related to pleiotropic effects of statins before surgery and their ability to modulate surgery-related post-operative inflammation (Tuuminen et al. 2015). Of note, Stein et al. (2012) showed that statin use could protect against both incident glaucoma and need for glaucoma medication, suggesting that the benefit of statin use might play out early during glaucoma development. Based on our findings, we can only speculate that statins might prevent inflammation and/or fibrosis of the anterior segment in the regions of conventional aqueous outflow (in TM and Schlemm's canal) (Ferrer 2006). Statins may also make the retinal tissues less vulnerable to operationrelated damage during vitrectomy, if used long-term before vitreoretinal surgery. Glaucoma can also result from obstruction of drainage structures in the anterior segment (in TM) due to post-vitrectomy complications such as rebleeding, the risk of which is highest among eyes that have undergone complex vitrectomies. Rebleeding can trigger rise in IOP, inflammation and fibrosis. Thus, systemic medication such as statins, administered longterm prior to surgery as well as during post-operative management, could be the key to success when evaluating the risk of glaucoma and other adverse effects such as reoperations and PVR formation after vitreoretinal surgery ). However, more exact associations of systemic statin therapy and development of glaucoma could not be investigated in our observational setting, since the number of patients undergoing vitreoretinal surgery was low.
The nested case-control design that was utilized in our current study is known to give good estimates for hazard rates that are approximated by odds ratios (Biesheuvel et al. 2008). However, like in all observational studies, it is not guaranteed that all pertinent risk factors are available to be used in statistical modelling. Therefore, it is possible that there is remaining confounding or other types of bias affecting the study results.

Conclusions
Our population-based study investigated the association of glaucoma with vitreoretinal surgery in Finland. Although real-world evidence may be subject to bias and confounding  factors, and despite the limitations of our study, we could show that the risk of any glaucoma is increased after most performed vitreoretinal surgical procedures. Interestingly, diabetes was not associated with the risk of glaucoma after vitreoretinal surgery. In addition, although exposure to statin in the longterm preoperative period was associated with reduced glaucoma risk after vitreoretinal surgery, no strong conclusions can be drawn from the subanalysis without information on the dose or type of statin. Further prospective research is thus warranted to elucidate the risk and clinical course of glaucoma after vitreoretinal surgery and in relation to novel antiinflammatory and anti-fibrosis agents.

Authorship
All authors have given their final approval of the version to be published; and SL and JH have given the agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. JH had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: SL, JH. Acquisition, analysis, or interpretation of data: JH, SL, AV. Drafting of the manuscript: SL, JH, AV. Critical revision of the manuscript for important intellectual content: SL, JH, AV, MH, EG, AK. Conflict of Interest Disclosures: JH reports funding paid to him from the University of Helsinki, Finland. No other disclosures were reported.