Analysis of retinal detachment after post-operative endophthalmitis treated with 23G Pars Plana Vitrectomy

Background


Abstract Background
To evaluate the rate, risk factors, functional outcome and prognosis in eyes with retinal detachment after post-operative endophthalmitis treated with 23G pars plana vitrectomy.

Methods
Electronic patient les from 2009 until 2018 were screened for the presence of an endophthalmitis. Included were 116 eyes of 116 patients. This population was evaluated for the rate of retinal detachment after 23G Pars Plana Vitrectomy for endophthalmitis following cataract surgery or intravitreal injection. The main outcome measures are retinal detachment and visual acuity.

Results
Reason for endophthalmitis was previous cataract surgery in 78 patients and following intravitreal injection in 38 patients. First clinical evidence of endophthalmitis was present in median 5 days after the triggering surgery. Twenty-ve eyes (21.55%) developed a retinal detachment in average 25 days after endophthalmitis. RD is signi cantly associated with preoperative visual acuity (p = 0.001).

Conclusions
Modern 23G vitrectomy technique seems not to lower the rate of retinal detachment after vitrectomy for endophthalmitis. And we also emphasize the prognostic role of preoperative visual acuity in RD development of the endophthalmitis treated with 23G pars plana vitrectomy. Background Endophthalmitis (EO) is a severe intraocular in ammatory response. It is typically divided into exogenous, endogenous (systemic infection in an immune-compromised patient), or masquerade syndromes (large cell lymphoma). Exogenous is mostly postoperative (e.g., cataract surgery), but may also be posttraumatic or related to organisms with an ability to penetrate intact corneas. It can be classi ed as either culture-positive or culture-negative (sterile) 1, 2 . and further strati ed into an acute form (within 6 weeks after surgery) which is the most common 3 and a delayed-onset form (more than 6 weeks after surgery).
Whatever form it may assume, EO is a serious and dangerous ocular condition and can be very challenging for the vitreoretinal surgeon because visibility can be severely compromised due to corneal edema, anterior chamber cells and non-transparent vitreous 4 . The toxins produced by the infecting pathogens and the resulting in ammatory responses can be destructive for the retina and lead to complications like retinal necrosis 4 or photoreceptor damage to the retina 5 .
Retinal detachment (RD) is a complication of both EO and the surgical procedures used in its treatment.
The rate of RD in the management of EO varies between 9 and 21% [6][7][8][9] . RD was related to capsular rupture, noxious bacteria and an early additional procedure in the Endophthalmitis Vitrectomy Study (EVS). It led to a poor visual prognosis, with 27% of patients achieving a nal best corrected visual acuity (BCVA) of 20/40 9 .
The approach to treatment of endophthalmitis is not consistently agreed on by vitreoretinal surgeons.
The main objectives of this retrospective multi-center study was to evaluate the rate and the risk factors of RD after surgical treatment of patients with severe acute exogenous postoperative endophthalmitis having no BCVA exclusion criteria and to contribute with a recommendation to the debate about what is the proper surgical strategy for these complicated cases.

Methods
In this retrospective study data of endophthalmitis patients from the departments of ophthalmology at the university clinic of Hamburg Eppendorf and the university clinic Eberhard Karls in Tübingen, Germany were evaluated.
Electronic patient les (Hamburg: IFA (ifa systems AG, Germany), Tübingen: Arzt-Informations-System (AIS)) were screened from 2009 until 2018 for the rate of EO. In both centers the study included cases which were initially treated in those clinics or operated elsewhere and referred for treatment. Included were patients with EO following cataract surgery and intravitreal injection who were treated with 23 G PPV and intravitreal medication (vancomycin 1 mg/0.1 mL and ceftazidime (2.225mg/ 0.1 mL), voriconazole (0.1 mg/0.2cc)). Patients with other reasons for EO (endogenous source, post trauma, post ltrating surgery and post PPV) and patients treated with VTB were excluded. Criteria employed to diagnose endophthalmitis were fundoscopy, ultrasound with vitreous body in ltration, pain, hypopyon, anterior chamber in ammation and medical history. Recorded parameters were patient related data, preexisting general health conditions, EO related data, BCVA and treatment. The population was further evaluated for the rate of RD after surgical treatment of endophthalmitis.
The vitrectomy did not include a peripheral shaving of the vitreous base and a posterior vitreous detachment (PVD) was not induced in any of the cases because posterior vitreous was already detached according to surgical reports. After completion of the vitrectomy a thorough examination of the peripheral retina was performed in order to locate any retinal breaks.
The study adhered to the tenets of the Declaration of Helsinki. The study was a retrospective data collection that was anonymized at the source. The study has been reviewed and approved by the ethics committee of Hamburg (PV7372) and written consent from the patients was not needed.

Statistical analysis
All analyses were conducted using statistical software. Association was tested using the Chi-Square Test.
Differences in time of factor variables were tested with McNemar Test. The distribution of quantitative variables was given as median. Twenty-ve (Q25%) and seventy-ve (Q75%) quartiles were calculated.
Statistical signi cance was set at p < 0.05.

Results
This retrospective study included 116 eyes of 116 patients with EO. Mean age was 74 years (range 48 to 96 years). 47 patients (40.52%) were male and 69 patients (59.48%) were female. Out of these 116 patients 19% were treated for diabetes, 62.9% for arterial hypertension and 4.3% were immunosuppressed. Clinical evidence of endophthalmitis was reported in median 5 days after the causing incident (Q25: 3 days, Q75: 7.25 days).
Reason for EO was previous cataract surgery in 78 patients and following intravitreal injection in 38 patients.
Surgery was performed at the same day of presentation in both clinics. All patients were treated with PPV in combination with intravitreal antibiotics.
An anterior chamber hypopyon was present in 81.9% of the patients.
RD occurred in 25 (21.55%) eyes with endophthalmitis after an average of 25.4 ± 16.8 days. RD is statistically signi cant associated with preoperative visual acuity (p = 0.001). There is a slight tendency to lower incidence for eyes with better visual acuity (Spearman correlation rho=-0.292, p = 0.001). For the distribution of RD by preoperative visual acuity see Table 1.  6 eyes (5,17%) were removed by enucleation due to phthisis bulbi. In these eyes no successful retinal reattachment was possible.
Ocular samples were obtained from vitreous sampling at the beginning of PPV. In 44.6% no growth was detected, in 50.7% gram + bacteria, in 3.1% gram -bacteria and in 1.6% fungal. No signi cant correlation was found between microbiological result and retinal detachment, p = 0.28.
There are no statistically signi cant differences between EO following cataract surgery and intravitreal injection. See Table 2.

Discussion
Infectious EO is an in ammatory reaction that poses a high risk of severe visual loss. During any intraocular procedure, prevention of EO should be a priority because of the multiple sources of contamination [10][11][12] .
A number of authors and studies addressed the problem of RD due to EO and its surgical treatment. In the EVS, the rate of postoperative RD was 7.8% in the 20-gauge vitrectomy subgroup. Later then, due to the advancement of surgical techniques and technology, re-evaluation of this study's results is needed 13,14 .
The of acute EO at the time of initial PPV or during follow up. Silicone oil proved to be effective in stabilizing the retina but the BCVA was poor in almost all patients due to the severity of the cases 15 . Previously, due to the fear of infection behind the silicone oil bubble, there had been a reluctance to use silicone oil as a tamponade agent for EO 26 . Later, silicone oil was proved to have an antibacterial and antifungal effect in vitro. The possible mechanisms of its antimicrobial activity that were reported are nutritional deprivation and toxicity 27 . The dosage of intravitreal antibiotics in eyes treated with silicone oil injection still remains controversial. Hegazy's study demonstrated a retinal toxicity in silicone oil-lled rabbit eyes, when the full dose of intravitreal antibiotics was used 28 . Nevertheless, those results still might not apply to the human eyes. Still we believe it is wise to reduce the dose of intravitreal drugs to about 25% of the dose that is usually injected because all intravitreal drugs will only distribute in the small aqueous phase surrounding the silicone bubble.
On the microbiological side, the results of the organisms identi ed in our study are in accordance with other studies and there was no statistically signi cant correlation between the microbiological ndings and the occurrence of RD or the initial BCVA (no statistical signi cance between Gram + bacteria and severity of EO, BCVA and rate of RD) 29,30 .
The treatment strategy of a severe EO is complicated and there is no clear protocol. Of essence is time and the goal is to evacuate the infection and administer antibiotics. The intravitreal injection of antibiotics and the vitrectomy are the standard and main therapeutic options. Every option has advantages and disadvantages. While vitrectomy allows the as complete as possible evacuation and removal of the infection, it is often not possible, since vitreoretinal surgeons and vitreoretinal operating rooms are relatively fewer. The VTB and intravitreal antibiotics injection have their own advantage. For example, they offer a smaller sample, permit earlier intravitreal antibiotics injection and microbiology tests 31 . Vitrectomy has evolved over the years after the EVS study (smaller gauges, faster surgical procedures, minimal invasive) but the rates of RD vary and can be still high in severe cases of EO as demonstrated not only from the EVS study but from other authors using modern PPV techniques (23 and 25 gauge systems) 15,17,18,25 .
The high rate of RD in our cohort cannot be attributed to iatrogenic intraoperative breaks or to the vitreous sampling. No shaving of the vitreous base was performed (in combination with detailed indentation search of the peripheral retina) and the posterior hyaloid was not actively detached because it was already detached in all cases. We also did not use for vitreous sampling undiluted vitreous but diluted.
Chiquet et al have reported that undiluted vitreous sampling at the start of PPV leads to hypotony, with a potential risk of vitreoretinal tractions, haemorrhages and RD. This can be avoided using diluted samples, since both samples have the same microbiological e ciency using PCR 32 .
In our study, retinal detachment is statistically signi cant associated with preoperative visual acuity, which is similar to the ndings of Doft et that RD is more likely to develop in patients who have the most severe presentation with visual acuity of LP only 9 . On the other hand, Chiquet et reported that other risk factors for RD in patients who had a vitrectomy after cataract surgery were diabetes and vasculitis 16 . Our study could not nd a statistical signi cant correlation between microbiological ndings (especially Gram + bacteria), diabetes mellitus, immunosuppression and severity of EO, and RD rate. Vitrectomy offers the advantage of as complete as possible evacuation of the infection but is associated with a spectrum of complications like RD.
One of the limitations of most of the studies in the literature today dealing with this very complex problem are the retrospective nature, lack of a de ned treatment protocol, treatment by multiple vitreoretinal surgeons and exclusion of cases due to the complexity of the disease and poor visual prognosis of this condition. These facts apply to our study also, but the large number of cases, the inclusion of two retinal centers, gives us optimism that our conclusions could shed some light on this complex issue.

Conclusions
The ndings of this study suggested that modern 23G vitrectomy technique seems not to lower the rate of RD after vitrectomy for endophthalmitis. The risk of retinal detachment still remains high in spite of the updated vitreoretinal techniques, especially with a higher cutting rate. And we also emphasize the prognostic role of preoperative visual acuity in RD development of the endophthalmitis treated with 23G pars plana vitrectomy, presumably due to the in ammatory effect on the vitreous and retina. The study adhered to the tenets of the Declaration of Helsinki. The study was a retrospective data collection that was anonymized at the source. The study had been reviewed and approved by the ethics committee of Hamburg (PV7372) and the need for written consent from the patients was waived by the ethics committee of Hamburg (PV7372).

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interest
The authors declare that they have no competing interests.

Funding
There was no funding received in relation to this article.

Authors' contributions
Skevas C designed the research; Dimopoulos S and Bartz-Schmidt KU collected the data; Spitzer M.S. revised the manuscript; Zheng Y and Casagrande M wrote the manuscript. All authors read and approved the nal manuscript.