Factors related to postoperative vitreous hemorrhage after small-gauge vitrectomy in proliferative diabetic retinopathy patients

Proliferative diabetic retinopathy (PDR) is the leading cause of blindness in working-age individuals [1]. Pars plana vitrectomy (PPV) may be required to treat the complications of PDR, such as vitreous hemorrhage (VH) and tractional retinal detachment (TRD), to prevent patients from severe, irreversible vision loss. Postoperative vitreous hemorrhage (POVH) is a common PPV complication and occasionally requires reoperation.

PDR, as a complication of diabetes, is often combined with diabetic cardio-cerebral vascular diseases (CVDs) or chronic kidney diseases (CKDs) [2‐7]. Systemic diabetic complications may require long-term anticoagulation or antiplatelet therapy to prevent thromboembolism. The surgeon often must balance the risk of thromboembolism and the risk of bleeding in the perioperative adjustment of anticoagulation and antiplatelet therapy. The perioperative guidelines on the adjustment of anticoagulation and antiplatelet therapy in noncardiac surgery show that in acute coronary syndrome, for patients with percutaneous intervention or drug-eluting stents, aspirin is recommended to be continued and clopidogrel should be restarted as soon as possible after surgery [5]; in patients with atrial fibrillation a mechanical heart valve, or venous thromboembolism, warfarin is typically stopped five days before surgery and resumed within 24 h postprocedure in cases of moderate bleeding risk, and low-molecular-weight heparin (LMWH) bridging therapy may be considered in the perioperative period [8]; in patients with CKD, heparin is recommended. The LMWH bridging therapy is commonly used in dialysis patients or diabetic CKD patients during the perioperative period [3]. There is no guideline or consensus on the perioperative adjustment of anticoagulation or antiplatelet therapy in diabetic vitrectomy. The consensus regarding ocular surgery recommends that there is no need to stop anticoagulation therapy in only cataract surgery, which is regarded as a procedure with a minimal bleeding risk [9].

In the 20-gauge vitrectomy era, without intravitreal anti-vascular endothelial growth factor (IV anti-VEGF) agents, endodiathermy, or intraoperative photocoagulation, the POVH occurrence in diabetic vitrectomy was relatively high, 39–63% [10, 11], and was related to the continued use of antiplatelet or anticoagulant agents therapy during vitrectomy [12]. However, in the small-gauge vitrectomy era, with the use of IV anti-VEGF agents, intraoperative photocoagulation, and endodiathermy, POVH occurrence has decreased dramatically to 3.1–18.9% [13‐17]. It is believed to be unrelated to the continued use of anticoagulants in dialysis patients [18] or antiplatelet therapy [12, 14, 18, 19]. In addition, a survey on the consideration of discontinuing anticoagulation or antiplatelet therapies during vitrectomy shows that only 9% of surgeons feel that they would stop anticoagulation if possible for diabetic vitrectomy and retinectomy [20].

Extensive vitreoretinal adhesion and dissection of the fibrovascular proliferative membrane (FVM) are known risk factors for POVH [13, 17, 21, 22]. Additionally, the previous studies mentioned above did not deal with patients with extensive FVM cases, and data about the relationship of POVH to anticoagulation and antiplatelet therapies in patients with severe PDR in the small-gauge vitrectomy era are limited. Thus, we examined a series of PDR patients, including those with severe cases with extensive FVM and TRD, investigated the relationship of POVH to anticoagulation and antiplatelet therapies, and sought to address these research gaps [21, 22].

This retrospective study included a cohort study of consecutive hospitalized patients with PDR who underwent PPV at our hospital during 1.1.2016–12.31.2017. If there was bilateral involvement, the eye with more severe ocular manifestations was selected. This study was approved by the Ethics Committee of Beijing Tongren Hospital, and it adhered to the tenets of the Declaration of Helsinki. The requirement for informed consent was waived.

The inclusion criteria were as follows: (1) PDR patients with TRD or VH who were hospitalized for PPV, (2) patients with available medical and operative documentation and laboratory tests to confirm diabetic systemic complications; (3) patients with details on the perioperative adjustments of anticoagulation and antiplatelet therapies; (4) patients with at least a three-month follow-up record. Exclusion criteria included: (1) VH which was found to be unrelated to PDR during PPV; (2) PDR eyes that underwent PPV for an epiretinal membrane or macular hole; (3) eyes with a history of PPV; (4) a lack of information on the history of DM, hypertension (HTN), evaluation on DM-related CVD, or CKD; (5) a lack of information on the use of anticoagulation, antiplatelet medication, or operative records; or (6) a lack of at least 3-month follow-up data.

Demographics, including age; gender; preoperative characteristics, including DM duration and DM medication, and the history of diabetes-related complications, including stroke [23] coronary artery disease (CAD) [23], congenital heart failure, diabetic foot and CKD [24], [25], and HTN [26]; anticoagulation or antiplatelet medications; and ocular characteristics, including the history of previous photocoagulation, cataract phacoemulsification extraction, IV anti-VEGF agents, visual symptom duration, visual acuity (VA), lens status, and the presence of iris neovascularization and neovascular glaucoma, were reviewed from the patients’ charts.

All the patients included in our study underwent preoperative anesthetic assessment. Physicians treated each patient with abnormal systemic conditions detected during preoperative screening. After the surgeon and the physician balanced the systemic risk of stopping anticoagulation or antiplatelet therapy and the benefit of PPV, the previous therapy was adjusted or left unchanged. The adjustment or maintenance of antiplatelet or anticoagulation medications in the perioperative period was recorded. The surgeon confirmed the restart of previous anticoagulation or antiplatelet therapy after PPV based on the post-PPV ocular signs for active bleeding. Adjustments to the renal dialysis plans and antiplatelet or anticoagulation therapy were recorded. PPV was postponed in patients who had had percutaneous transluminal coronary angioplasty, coronary artery bypass grafting, or cerebrovascular event in recent six months, especially those who were using dual antiplatelet therapy or had drug-eluting stent implants.

The patients had retrobulbar anesthesia or general anesthesia. All patients underwent 23/25 gauge 3-port PPV (Constellation Vision System, Alcon, Fort Worth, Texas; USA) (Stellaris PC, Bausch & Lomb, USA) with 5000 cuts/min vitreous cutting rates. Patients with active fibrovascular epiretinal proliferative membranes (FVMs) or severe VH obscured the fundus were treated with IV ranibizumab (IVR) within seven days before PPV. Pre-PPV IVR was recorded.

After a core vitrectomy, the extent of the FVM, the presence of macular-involving TRD, and photocoagulation scars were recorded.

Triamcinolone acetonide-assisted PPV was performed. Anterior-posterior vitreoretinal traction was released as much as possible. The induction of posterior vitreous detachment, FVM dissection, segmentation, delamination, endodiathermy, and the drainage of subretinal fluid, retinectomy, and intraocular tamponade was performed according to each patient’s particular need. Peripheral vitrectomy with scleral indentation and endolaser photocoagulation was performed for each patient. Intraocular pressure elevation, perfluorocarbon liquids, and endodiathermy were used to handle the intraoperative bleeding.

The severity of intraoperative bleeding was recorded as follows:

grade 0: none;

grade 1: minor bleeding stopping spontaneously or with transient bottle/pressure elevation;

grade 2: moderate to severe bleeding requiring endodiathermy or with the formation of broad sheets of clots [27].

Incomplete scatter photocoagulation was defined as a lack of preexisting photocoagulation in all 4 quadrants or needing more than 500 laser spots during PPV. The operation time was calculated from when the first PPV trocar incision was made to removing the eyelid speculum after PPV. The use of additional procedures, including cataract extraction, silicone oil tamponade, photocoagulation, and endodiathermy was recorded. The laser points were recorded.

The extension of FVM was recorded as follows:

grade 0: the absence of any adhesion;

grade 1: multiple point adhesions with or without one broad adhesion (broad adhesion was defined as focal adhesion at three sites or more);

grade 2: 1–3 broad adhesions posterior to the equator;

grade 3: 3 + broad adhesions posterior to the equator or 2 or fewer in-quadrant with adhesions anterior to the equator;

grade 4: broad adhesions anterior to the equator at multiple sites [28, 29].

FVMs were classified as predominantly neovascular, mixed neovascular and fibrotic, and predominantly fibrotic [27].

This was a retrospective cohort study, and the role of POVH-related factors needs further exploration in a prospective cohort. We did not analyze patients with spontaneously resolved POVH when we investigated the factors related to POVH. Previous work has shown that postoperative IV-anti-VEGF agent administration effectively resolves VH [21, 45]. We did not include the details on the management of POVH before the re-PPV. We failed to obtain postoperative VA data in this cohort.