In our study, postoperative CST and MV were significantly higher after GDD than after trabeculectomy in our study (p ≤ 0.039 respectively). Also, CME occurred nearly four times more frequently after GDD implantation compared to trabeculectomy, this difference was statistically significant.

Postoperative CME occured in 4.8% of eyes undergoing trabeculectomy with MMC. This frequency seems consistent with existing evidence for CME after trabeculectomy and comparable to the incidence of CME after cataract surgery. Previous studies described the incidence of postoperative CME to range between 3 and 9% after trabeculectomy with MMC [11,12,13,14] and between 4 and 20% for cataract surgery [9, 10]. Differences in study design, length of follow-up or inclusion criteria might explain the relevant range of CME incidence rates in these studies. Some comprised of a twelve months follow-up while others included a follow-up of up to 5 years. The risk of CME occurrence after trabeculectomy has been linked to the type of glaucoma. Manabe et al. found pseudoexfoliation glaucoma to increase the risk of CME compared to POAG [14]. Also, the number of previous surgical procedures and comorbidities varied between different study cohorts.

The risk for postoperative CME seems to be relevantly higher than after non-penetrating glaucoma surgery like 360° suture trabeculotomy or canaloplasty. In those surgical procedures a recent study found postoperative CME to occur after 0.7 and 0.3% of stand alone surgeries respectively [19].

Eyes undergoing GDD surgery had a CME rate of 19.4% in our study. This matches existing evidence, where rates of CME formation between 3 and 22% after GDD implantation have been reported [11,12,13]. In a retrospective study on 185 glaucomatous eyes, Bhakta and colleagues described a rate of 22% for occurrence of visually significant CME after GDD implantation. This rate was significantly associated with the number of previous intraocular surgeries [13]. This study included eyes with inflammatory ocular disease (i.e. iritis) and was restricted to eyes with the occurrence of visually significant CME only.

The mechanisms of postoperative formation of CME have not yet been fully understood. A sterile inflammation resulting an insufficiency of the blood–aqueous barrier is discussed [6]. Surgical trauma seems to lead to a release of inflammatory mediators as well as free radicals [20]. This leads to a breakdown of the blood-aqueous barrier, which is indicated by an elevated level of laser flare in the anterior chamber in eyes with postoperative CME [21]. As a result, the vascular permeability is increased, and fluid accumulates in extracellular spaces of the perifoveal retina [20]. Many other factors have been described to be associated with a higher risk of postoperative CME, these include a history of uveitis [22], the history of previous pars plana vitrectomy for retinal detachment as well as the preexistence of epiretinal membranes (ERM) [23]. Also topical prostaglandin analogs may promote macular edema [24, 25]. While the risk of postoperative formation of CME seems not increased due to glaucoma alone [26] and also ab interno glaucoma surgery combined with cataract surgery does not seem to have an elevated risk compared to stand alone cataract surgery [27], there is evidence for an increased risk of postoperative formation of CME after GDD surgery to treat glaucoma [12]. Also the type of glaucoma seems to have an influence on the rate of formation of CME as Manabe et al. reported significantly higher rates of CME after trabeculectomy in pseudoexfoliation glaucoma than in POAG [14]. The frequency of CME formation shows some discrepancy between real life data and data from clinical trials. [12, 13]

The larger surgical trauma and an increased inflammatory response to the medical implant could be two reasons explaining the increased risk for CME development in GDD surgery as well as significantly higher CST after GDD.

Prior to surgery, 91% of eyes were treated with prostaglandin analogs in our study (95% before trabeculectomy, 87% before GDD). Potentially this could also have increased the risk of postoperative CME compared to surgery in non-glaucomatous eyes. We found no significant difference in the occurrence of CME between eyes, which were previously treated with prostaglandin analogs and eyes, which were not. This is in line with a recent study by Fakhraie et al. who found no increased risk for postoperative CME after cataract surgery in patients treated with latanoprost [28, 29].

Postoperative CME can lead to a significant reversible or non-reversible loss in visual acuity [30].

Interestingly, recently our group reported that intense, early postoperative topical steroid therapy could significantly reduce the CME risk after posterior lamellar grafting combined with phacoemulsification [31]. In eyes receiving prednisolone eye drops hourly for the first postoperative week the CME rate was 0% versus 12% in eye just receiving 5x/day (p < 0.05). That potentially opens a new avenue for prevention of CME also in eyes after glaucoma surgery [31].

Limitations of our study include the retrospective set up of the study. The number of study eyes and strict inclusion and exclusion criteria reduce this effect.

Due to the strict exclusion criteria only 73 patients out of 535 screened patients could be included in the study. This was mainly due to other glaucoma diagnoses, but also due to missing preop or postop SD-OCT. This could depict a potential bias, as those patients with full visual acuity not always receive a SD-OCT. Also, patients with postoperative complications are referred more often to the tertiary center for follow-up. These two factors could have led to an overestimation of CME in our cohort. This applies to both groups and therefore should not bias the direct comparison on the frequency of CME occurrence.

Due to the use of real life data, our study falls short to allow a clear differentiation between the variables number of previous surgeries and type of surgical procedure regarding their impact on CME occurrence and on quantitative OCT measurements.

A logistic regression model did not reach the threshold for statistical significance; however, a tendency towards a higher relevance of type of surgery may be deducted from the variable’s higher odds ratio and 95% confidence interval. Furthermore, we did not find a significant correlation between the number of previous intraocular surgeries and the occurrence of a CME (p = 0.235). The conditions of this study represent conditions in real life, where patients receiving GDD often have experienced a greater number of previous surgeries. To differentiate further between the influence of the number of previous surgeries and the type of surgical procedure, a prospective study would be necessary including only patients who receive trab or GDD as primary surgical intervention for glaucoma.

In conclusion, CME is a relevant complication after both trabeculectomy and GDD surgery. GDD surgery as primary surgical option for non-uveitic open angle glaucoma is discussed controversially. At our center we use GDD only as treatment option for eyes with high risk of failure or previously failed trabeculectomy with MMC.

This study found, that in real life CME occurred significantly more often after GDD in patients with a higher number of previous surgeries than after trab in patients with a lower number of previous surgeries. This may represent a relevant information for the choice of surgical treatment and preoperative risk assessment.