Clinical analysis of persistent subretinal uid after pars plana vitrectomy in macular involving diabetic tractional retinal detachment

The purpose of this study was to analyze the duration of persistent subretinal uid (PSF) and contributing factors after pars plana vitrectomy in patients accompanying macular involving diabetic tractional retinal detachment (TRD). We included 33 eyes of 33 patients diagnosed macular involving diabetic TRD and received pars plana vitrectomy between 2014 and 2019 were reviewed retrospectively. PSF was conrmed on optical coherence tomography. The duration of PSF after surgery and relevant ocular and systemic factors according to the duration of PSF were analyzed. The prevalence of PSF was 81.8% a 1 month, 54.5% at 3 months, 36.4% at 6 months and 12.1% at 12 months after surgery and mean duration of PSF was 5.2 ± 4.9 months. The mean best corrected visual acuity was signicantly improved after 1 year and the nal visit than baseline. Blood urine nitrogen, creatinine, and estimated glomerular ltration rate (eGFR) were signicantly associated with duration of PSF in univariate analysis, but only eGFR was signicantly associated in the multivariate analysis. Therefore, patients with impaired kidney function tend to have delayed subretinal uid absorption and it would be necessary to explain in advance that PSF could be long-lasting in patient with impaired kidney function. Values are presented as the mean ± standard deviations.BUN: blood urine nitgogen; eGFR:estimated glomerular ltration rate; HbA1c: Hemoglobin A1c; ILM: internal limitingmembrane; PSF: persistent subretinal uid; SO: silicone oil; VEGF: vascular endothelialgrowth factor


Introduction
The prognosis of proliferative diabetic retinopathy (PDR) has been greatly improved in the era of panretinal photocoagulation and anti-vascular endothelial growth factor (VEGF) intravitreal injection 1,2 .
However, tractional retinal detachment (TRD) is one of the most feared complications of PDR and indications of pars plana vitrectomy in PDR.
Pars plana vitrectomy for TRD still may be one of the most complex vitreoretinal procedures. The goals of surgery are to remove vitreous hemorrhage, reduce anterior to posterior vitreous traction on the retina, and to relieve epiretinal tangential traction on which proliferative tissue grows. However, iatrogenic retinal breaks are the main complication of surgery and intraocular tamponade is used to repair retinal breaks.
Persistent subretinal uid (PSF) was detected by optical coherence tomography (OCT) after repair of rhegmatogenous retinal detachment [3][4][5] , and it was associated the chronic nature of uid with higher viscosity, protein composition and cellularity 6 . Recently, clinical cases of PSF were reported after vitrectomy for diabetic TRD 7 . Karimov et al. 8 analyzed the prevalence of PSF in macular involving diabetic TRD after vitrectomy. However, all of cases received pars plana vitrectomy with intraocular tamponade, and there was no published report to our knowledge to nd out the effect of intraocular tamponade on PSF in diabetic TRD surgery.
The purpose of this study is to report the prevalence of PSF and to analyze ocular and systemic factors contributing to duration of PSF in patients with PDR accompanying macular involving TRD after pars plana vitrectomy.

Demographics of patients
Ninety patients were initially included; however, 57 patients were excluded according to the exclusion criteria. Thus, a total of 33 patients (33 eyes) were enrolled in this study. The demographic and clinical characteristics for enrolled patients are presented in Table 1. Among the 33 patients, 20 were male, and 13 were female. The mean age of all patients was 49.9 ± 12.2 years and mean follow up duration was 30.0 ± 18.8 months. The duration of diabetes mellitus was 10.0 ± 6.4 years, 13 patients had hypertension, and 8 patients had chronic kidney disease on hemodialysis.
HbA1c was 9.5 ± 2.3, blood urine nitrogen (BUN) was 24.9 ± 11.6 mg/dL, creatinine was 1.4 ± 0.9 mg/dL, and estimated glomerular ltration rate (eGFR) was 64.2 ± 30.0 mL/min/1.73m 2 at preoperative laboratory results. 27 eyes had received previous panretinal photocoagulation and 7 eyes received anti-VEGF treatment prior to surgery. 26 eyes were underwent preoperative intravitreal bevacizumab injection in a day before vitrectomy and 24 eyes received combine surgery of phacoemulsi cation. During pars plana vitrectomy, ILM peeling procedure was performed in 6 eyes. 25 eyes had preexisting or iatrogenic retinal break during vitrectomy and SO tamponade was carried out after uid-air exchange. Among the 25 eyes, intraoperative SRF drainage was conducted from a retinotomy site in 20 eyes by active aspiration.
SO-lled eyes were underwent secondary SO removal surgery at 10.2 ± 5.5 months (range, 3.4 to 22.8 months) after primary surgery.

The duration of PSF
The mean duration of PSF was 5.2 ± 4.9 months (range, 0.2 to 20.5 months). The prevalence of PSF was 81.8% at 1 month, 54.5% at 3 months, 36.4% at 6 months and 12.1% at 12 months on the macula OCT scans after surgery. Figure 1 was shown a representative case of PSF on the OCT scans. A Kaplan-Meier graph was used to display the estimated survival probability of PSF for patients, as shown Fig. 2.
Clinical factors associated with the duration of PSF Univariate and multivariate linear regression analyses of associations between clinical factors and the duration of PSF was presented in Table 2  There is controversy over intraoperative procedures such as internal drainage or intraocular tamponade during pars plana vitrectomy in TRD. Meredith et al. 10 proposed the technique of membrane segmentation in TRD surgery and stated that there is no need to drain the SRF because relief of traction is su cient to allow reattachment of the retina and intraocular tamponade is not necessary if the traction is surgically released without creating retinal breaks. But, in fact, intraocular silicone oil or gas was frequently used after pars plana vitrectomy for severe TRD cases 10,12−14 .
The authors focused on whether intraocular tamponade or SRF drainage procedure affects the duration of PSF. Karimov et al. 8 reported the duration of PSF in diabetic TRD after vitrectomy that eyes with SO tamponade showed signi cantly faster SRF absorption and intraoperative drainage showed also faster SRF absorption, but not statistical signi cant. As a result, they suggested that SO was a favorable option of tamponade in cases with intraoperative breaks and internal drainage. However, our result shows that intraocular SO tamponade and drainage of SRF during surgery are not related to the duration of PSF (Table 2).
In addition, we found that lower eGFR level is a signi cant risk factor of long-standing PSF. Basically, absorption of SRF depends on passive diffuse and active pumping by retinal pigment epithelium (RPE) 15 . However, in diabetic retinopathy, presence of SRF may be caused by impaired RPE pumping and disruption of external limiting membrane, which serves as a barrier to subretinal space and contributes uid shifting from intraretinal space to outer retina 16 . In addition, diabetic retinopathy with chronic kidney disease, decreased serum albumin may lower the intravascular osmotic pressure and increase hydrostatic pressure in outer retina or choroid that could leads to uid retention and ow into the subretinal space 17,18 . Therefore, decreased kidney function could be contributed to delayed absorption of SRF.
The present study has some limitations. This is retrospective study on a relatively small sample size.
Further researches on a greater number of cases with evaluation of characteristics of the PSF after TRD surgery will be necessary. Nevertheless, this study has signi cance in analyzing the risk factors of PSF in diabetic TRD patients after surgery.
In conclusion, we have shown that PSF is detected by OCT after successful diabetic TRD surgery. In addition, patients with impaired kidney function due to diabetic nephropathy tend to have delayed SRF absorption. Therefore, the investigation of preoperative systemic conditions on PDR patients should be considered before TRD surgery and it would be necessary to explain before surgery that PSF could be long-lasting in patient with impaired kidney function.

Methods
Medical records were reviewed after approval of the Institutional Review Board of Kyungpook National University Hospital (IRB No. 2021-02-009) and waived the requirement for informed consent because of the retrospective nature of the study. The research was conducted in accordance with the tenets of the Declaration of Helsinki.

Patients
We included patients who was diagnosed PDR with macular involving TRD and received pars plana vitrectomy from January 2014 to December 2019. The study included only the eye that had PDR with macular involving TRD and showed subretinal uid (SRF) on preoperative OCT. If the patient who had macular involving TRD in both eyes, we included only one eye with more poor visual acuity. All patients showed successful anatomical results after vitrectomy and completed at least 3 months follow up period with OCT examinations. Patients were excluded according to the following criteria: (1) diabetic TRD without macular involving; (2) diabetic TRD without SRF on preoperative OCT scan of the macula; (3) the presence of vitreous haze or hemorrhage that are not suitable for OCT imaging.
Ophthalmic examinations were performed including best corrected visual acuity (BCVA) using a Snellen chart, intraocular pressure (IOP) measurement, slit lamp examination, fundus examination, and OCT examination. All examinations were repeated at baseline and at every follow up periods after vitrectomy.
BCVA were converted to the logMAR (logarithm of the minimum angle of resolution) for statistical analyses. The detachment of macula was con rmed by a preoperative macular OCT scan. The volume mode scan of 6 × 6 mm area was performed using the spectral-domain OCT (Spectralis®, Heidelberg Engineering, Heidelberg, Germany). The presence of SRF was de ned on OCT image as a hypore ective space between the photoreceptor layer and the retinal pigment epithelium. PSF de ned as SRF present within any area on 6 × 6 mm OCT scans that persists after vitrectomy.
The surgical technique A microinvasive pars plana vitrectomy was conducted using 23 or 25 gauge instrument. When a combined surgery was planned, phacoemulsi cation of the cataract with the implantation of the intraocular lens was conducted at the beginning of the surgery. Vitrectomy was performed using the Constellation surgical system (Alcon, Forth Worth, Texas, USA) and included removal of the posterior vitreous and shaving of peripheral vitreous body, peeling of the posterior hyaloid, neovascular, and brous epiretinal membranes from the retinal surface, and endolaser photocoagulation. Internal limiting membrane (ILM) peeling was carried out after staining with indocyanine green (ICG) dye using membrane forceps, if epiretinal membranes with a risk of postoperative re-proliferation was presented. Internal drainage of SRF was carried out if a preexisting or iatrogenic retinal break was observed intraoperatively. For silicone oil (SO) tamponade, if necessary, 5700 centistokes silicone oil (Oxane®, Bausch and Lomb, USA) was injected to the posterior surface of the pupil after uid-air exchange.

Statistical methods
Statistical analyses were performed using the Statistical Package for the Social Science software 20 (IBM Corp., Armonk, NY, USA). A plot of Kaplan-Meier graph was showed survival time of PSF. Repeated measures analysis of variance corrected by the Bonferroni method was performed to compare the mean BCVA for the follow-up periods. The Pearson correlation coe cient was calculated the mean BCVA at the nal visit and the duration of PSF. Ocular and systemic factors associated with the duration of PSF were analyzed using univariate and multivariate linear regression analyses. P value < 0.05 was considered signi cant for all statistical tests. Kaplan-Meier graph illustrating the survival probability of persistent subretinal uid for patients.