Inflammation should be controlled preoperatively, as large amounts of preoperative inflammatory activity have been significantly linked with increased postoperative inflammation in patients with uveitis undergoing vitrectomy (p < 0.02) [15]. Patients may be placed on high-dose systemic oral steroids (i.e., 1 mg/kg prednisone) starting 1–2 days prior to surgery in order to minimize the risk of increased postoperative inflammation. A standard three-port vitrectomy is employed. Smaller gauge techniques may help to minimize increased iatrogenic inflammation, but 20-gauge vitrectomy may facilitate concomitant procedures such as pars plana lensectomy. If endolaser is performed, care must be taken, as excessive laser may also exacerbate postoperative inflammation by increasing vascular permeability [16, 17].

Complete removal of the vitreous may improve long-term outcomes [18]. Triamcinolone may be used intraoperatively to visualize and facilitate complete removal of vitreous from the macular interface as well as the anterior hyaloid [19, 20]. Upon completion of the core vitrectomy, triamcinolone is injected into the vitreous cavity, highlighting the residual vitreous cortex and hyaloid. A silicone-tipped needle or internal limiting membrane (ILM) forceps may be used to remove adherent posterior hyaloid. Triamcinolone may be left in the vitreous cavity to decrease postoperative inflammation [21].

The role of removal of the ILM for uveitic ME remains unclear. In their report on vitrectomy of 42 eyes, Wiechens et al. (2001) removed the ILM in 15 eyes. While ME improved in 60 % of eyes overall, there was no difference between eyes with and without ILM removal [22]. Gutfleisch et al. (2007) studied 19 patients with uveitic ME who underwent vitrectomy with ILM peeling and concurrent intravitreal triamcinolone [23]. Postoperatively, angiographic ME improved in 58 % of patients. The authors noted, however, that ILM peeling in uveitic ME is difficult and may cause tissue damage; they do not recommend ILM peeling unless there is VMT present. Cho and D’Amico’s series (2012) of 24 patients with chronic ME undergoing 25-gauge vitrectomy with ILM peel included four with uveitis [24]. There was slight nonsignificant improvement in visual acuity and macular thickness as measured by spectral domain optical coherence tomography (SD-OCT) in these patients (429 μm to 407 μm; p = 0.92).

For diagnostic vitrectomy, as much undiluted vitreous should be obtained as possible to send for laboratory evaluation including cultures, PCR testing, cytology, and flow cytometry. Approximately 1 cc of undiluted vitreous may be safely obtained by standard vitrectomy techniques. Additional undiluted vitreous can be obtained via techniques such as infusion of air or heavy liquids. Using perfluoron as originally described by Quiroz-Mercado for large-volume vitreous biopsy [25], one may safely obtain 3–5 cc of undiluted vitreous [26].

While there are no large, randomized, controlled, prospective studies of vitrectomy for uveitic ME, since the early studies referenced above, there have been numerous small studies suggesting potential benefit. Becker and Davis reviewed 44 articles on vitrectomy for uveitis from 1981 to 2005, which included a total of 1762 eyes [27]. Overall, there was a reduction in the median percentage of eyes with ME from 36 % preoperatively to 18 % post-vitrectomy. Improvements in visual acuity and inflammation control were also suggested, but the highest evidence grade of the articles reviewed was CII-3: “at least fair evidence that the service can improve health outcomes but … the balance of benefits and harms is too close to justify a general recommendation” [28].

To date, there have been two small randomized, controlled, prospective trials comparing the effect of vitrectomy versus medical therapy in uveitic ME. Tranos et al. (2006) studied 23 patients with recalcitrant uveitic ME and no other macular pathology who were randomized into a vitrectomy group and a systemic medical therapy group [29]. Vision improved by two or more lines in 50 % of the eyes undergoing vitrectomy versus 18 % in the medical group. Angiographic improvement of ME was observed in 33 % of the vitrectomized eyes compared to 14 % in the control group. Quinones et al. (2010) randomized 20 eyes with recalcitrant intermediate uveitis to vitrectomy or immunomodulatory therapy; three eyes in each group had ME [30]. All three in the surgical group showed resolution of ME, while two of three in the medical group showed improvement in ME. In both of these small studies, differences did not reach statistical significance.

A number of reports have focused on the effect of vitrectomy for uveitic ME in specific anatomic locations or diagnoses. In 1992, Kaplan suggested that vitrectomy may provide an alternative to systemic immunosuppression in intermediate uveitis [31]. Since then, intermediate uveitis has been the most common diagnosis in reports of vitrectomy for uveitis, with generally positive results for both macular edema and vision [27].

Following vitrectomy for refractory uveitic ME in 53 patients, Wiechens et al. (2003) reported a resolution or reduction in ME in 59 % of patients with intermediate uveitis, 57 % with juvenile idiopathic arthritis (JIA) associated uveitis, and 41 % with multifocal choroiditis. There was a concurrent two-line improvement in Snellen visual acuity in 50 %, 71.4 %, and 41.7 %, respectively [32]. In a later report focusing predominantly on juvenile intermediate uveitis, ME was significantly reduced in 8 of 10 eyes post-vitrectomy [33].

In sarcoid uveitis, Kiryu et al. (2001) demonstrated resolution of medication-resistant ME in 14 (78 %) of 18 patients who underwent vitrectomy [34]. Half of the eyes also had peeling of ERM or adherent posterior vitreous cortex. Interestingly, in both the vitrectomy only and the membrane-peeling groups, seven of nine eyes showed improvement in ME.

Sullu et al. (2005) performed vitrectomy for posterior segment complications of Behçet’s disease in 20 eyes, including five with ME [35]. They noted complete improvement of ME in three eyes (60 %) after vitrectomy.

Llorenç et al. (2011) showed improvement in ME following vitrectomy in 16 eyes with human leukocyte antigen HLA-A29-positive birdshot chorioretinopathy [36]. Nine eyes had preoperative ME with a mean macular thickness of 537.8 μm by HD-OCT. Four of the nine eyes also had ERM; all nine had either ERM peeling or Brilliant Blue-assisted ILM peeling. ME improved postoperatively in eight of nine eyes (89 %), with a final mean macular thickness of 218.7 μm (p = 0.0039).

Patients with chronic uveitis and ME may develop fixed retinal cysts, enlarged foveal avascular zones, and/or thinning of intraretinal layers. Patients with such findings may not be good candidates for primary vitrectomy, as surgical management is unlikely to correct the underlying pathology.

The prevalence of ERM has been reported as 40–48 % in patients with uveitis [37–39]. Nicholson et al. (2014) reported on a large cohort of uveitis patients with ERM evaluated by SD-OCT [39]. Among the 598 patients, 246 had ERM in at least one eye. Multivariate analysis suggested that ERM is associated with approximately one line of visual acuity loss in these patients. Central retinal thickness of greater than 350 μm in conjunction with an ERM conferred a significant decrease in visual acuity compared to patients with central retinal thicknesses between 200 μm and 350 μm.

Results of ERM peeling in patients with uveitic ME have suggested some benefit. Dev et al. (1999) studied five eyes with ERM and ME diagnosed clinically or via angiography with chronic idiopathic pars planitis who underwent ERM peel with vitrectomy [40]. Four of five eyes had visual acuity improvement and reduction or elimination of ME. Kiryu et al. (2003) showed resolution of ME by fluorescein angiography (FA) in four of seven eyes with sarcoid uveitis that underwent ERM peel with vitrectomy [41]. Visual improvements, however, were not significant. Tanawade et al. (2014) showed improvement in vision in five eyes with ERM and concurrent uveitic ME and six eyes with concurrent ERM, VMT, and uveitic ME that underwent ERM peeling with or without ILM peeling [42]. Nine of eleven eyes showed resolution of ME and VMT on OCT at 3 months postoperatively.

The incidence of rhegmatogenous RD in patients with uveitis has been reported as high as 3 % [43]. As these are often complex retinal detachments, vitrectomy is usually indicated. There is little in the literature, however, regarding ME in uveitis patients undergoing vitrectomy for retinal detachment. Yu and Chung (1994) repaired seven traction retinal detachments (TRDs) and eleven combined traction/rhegmatogenous detachments in patients with chronic uveitis [44]. While the exact number was not reported, the authors did state that many of these eyes had preoperative ME. Postoperative ME was seen in two patients in the TRD group and one patient in the combined group. Recurrence of TRD occurred in two patients and combined RD in six patients.

A significant cause of decreased vision in uveitis may be non-clearing vitreous opacities and/or vitreous hemorrhage (VH). In a series of six eyes with pars planitis undergoing vitrectomy for VH, Potter et al. (2001) reported two eyes with preoperative ME [45]. Vision in both eyes improved; final acuity in one was 20/20, with the other 20/100 likely due to ME.

Ieki et al. (2004) performed vitrectomy for non-clearing vitreous opacity in 11 eyes, 5 of which had preoperative treatment-resistant ME [46]. After 6 months, all five eyes had either resolution or improvement of ME as determined by FA. Three of these eyes gained two or more Snellen visual acuity lines and achieved visual acuity of 20/40 or better, while the other two eyes had stable visual acuity at the final visit.

Vitrectomy has also been reported to reduce ME in juvenile uveitis with vitreous opacities. Trittibach et al. (2006) reported on 29 eyes that underwent vitrectomy for vitreous opacities (n = 25), VH (n = 3), and retinal detachment (n = 1) [33]. ME was reduced in eight of ten eyes that had preoperative ME (p = 0.021). Overall, LogMAR visual acuity improved from an average of 0.91–0.33 postoperatively (p = 0.001).

Combined cataract surgery and pars plana vitrectomy may be indicated if significant cataract is present, but the combined surgery may incite new or worsen preexisting ME. In 1979, Diamond and Kaplan studied 25 eyes that underwent combined vitrectomy and pars plana lensectomy without placement of intraocular lens (IOL). They reported resolution of preoperative cystoid macular edema (CME) in 4 of 12 eyes [47]. More recently, Androudi et al. (2005) reported 36 eyes with chronic uveitis that underwent combined phacoemulsification and pars plana vitrectomy [48]. Nine of the eyes (25 %) had preoperative ME confirmed by FA. Postoperatively, six of these nine eyes had persistent edema, while ten new cases of ME were identified. Only four of the ten new cases of ME resolved during the follow-up period.

Acceleration of cataract formation following vitrectomy for uveitis is seen in virtually all phakic patients, and significant cataract development has been reported as high as 100 % [34]. Other relatively common complications following vitrectomy for uveitic ME include increased or decreased intraocular pressure, RD, VH, and ERM. In an early series of 12 eyes undergoing vitrectomy for peripheral uveitis, Mieler et al. (1988) reported a 50 % repeat surgery rate for RD, VH, or cataract; nonetheless final acuity improved an average of five lines [49]. In their review of 44 papers assessing vitrectomy in a total of 1762 eyes, Becker and Davis counted postoperative complications including 112 progressing cataracts, 56 partial and 21 total RDs, 51 secondary glaucomas, 45 cases of hypotony and 15 of phthisis, 36 macular puckers, 22 VHs, 7 hyphemae, and 3 choroidal detachments [27]. In a retrospective review of 74 uveitic eyes that underwent 25-gauge vitrectomy, 56 of which had preoperative ME, Soheilian et al. found ERM formation in 23 %, elevated intraocular pressure in 11 %, irreparable RD in 6.7 %, subretinal neovascular membrane in 2.7 %, macular hole in 5.4 %, phthisis bulbi in 5.4 %, and chronic hypotony in 5.4 % [15]. Nicholson et al. found that ERM formed in patients with uveitis at a significantly greater rate after vitrectomy (16/141) compared with those that did not undergo vitrectomy (6/141) (p = 0.026) [39]. Vitrectomy with concurrent intravitreal triamcinolone was found to cause ocular hypertension in 9/19 (47 %) of patients [23].

Local or systemic corticosteroids, with or without concurrent systemic immunomodulating medications, may fail to resolve uveitic ME. Even when effective, local steroid injections often need to be repeated multiple times to achieve control of uveitic ME. The 0.59 mg fluocinolone acetonide (FlAc) intravitreal implant (Retisert; Bausch & Lomb) [50] releases steroid into the vitreous cavity for an average of 30 months and was approved by the FDA for treatment of noninfectious uveitis in 2005. Surgical implantation of a FlAc implant provides an alternative to multiple local steroid injections or systemic therapy.

Details of the implantation technique have been well described [51–53]. Briefly, the implant is prepared by securing a double-armed 8-0 prolene suture through the anchor strut of the implant with a single knot. After prepping the eye in aseptic fashion, a conjunctival peritomy is performed in an area with healthy appearing conjunctiva, away from underlying pathology such as a snowbank or traction. The surgeon may consider avoiding areas likely to be used for future glaucoma surgeries. Cautery is used to achieve hemostasis. In eyes that have previously undergone vitrectomy, an infusion line is placed to maintain intraocular pressure. A 20-gauge microvitreoretinal (MVR) blade is used to make a 3.5 mm full-thickness sclerotomy along a concentric line 4.0 mm posterior to the limbus. Any prolapsed vitreous may be cut with a vitrector or excised using a Weck-Cel sponge and Westcott scissors. The implant is inserted into the vitreous cavity with the drug-eluting portion facing anteriorly. The previously placed anchor suture is then passed through the sclera on either side of the incision and tied, thereby gently approximating the scleral incision. The tails of the double-armed prolene are then placed under interrupted 9-0 prolene sutures that are used to close the sclerotomy. The interrupted sutures are rotated to bury the knots. The proper position of the implant is confirmed by indirect ophthalmoscopy. Balanced salt solution is injected into the vitreous cavity to normalize the intraocular pressure. The conjunctival peritomy is closed with 6-0 plain gut sutures, and subconjunctival antibiotics are given.

Berger and Mendoza have suggested an alternate suture technique for closer approximation of the sclerotomy [54]. Slow-absorbing 8-0 polyglycolic acid sutures may be used on the inner aspects of the sclerotomy, prior to placing two distal 9-0 prolene permanent sutures. Use of fewer permanent sutures may also reduce the risk of conjunctival erosion.

The first report of long-term safety and efficacy of FlAc implantation to control posterior uveitis was published by Jaffe et al. in 2005 [51]. Thirty-six eyes were randomized to either a 0.59 mg or a 2.1 mg FlAc implant. With follow-up of at least 12 months in 72 % of patients, 24 months in 44 %, and 30 months in 25 %, only two patients were noted to have recurrent inflammation, both at 29 months or later. Visual acuity stabilized or improved in 90 % of patients, with the authors speculating that the improved vision was primarily due to reduced macular edema.

Results of the FlAc implant in a large group of patients were first reported by the multicenter Fluocinolone Acetonide Uveitis Study Group in 2006 [55]. Two hundred and seventy-eight eyes of patients with long-standing noninfectious posterior uveitis that had previously undergone systemic and local therapy were randomized to receive a FlAc implant of 0.59 mg or 2.1 mg and were subsequently followed for 3 years [56]. Patients with bilateral disease had FlAc implanted in the eye with the more severe uveitis.

Recurrence rates in the 0.59 mg dose group were reduced from 62 % over the year before implantation to 4 % at 1 year, 10 % at 2 years, and 20 % at 3 years after implantation. These recurrence rates were also significantly lower than those of the fellow non-implanted eyes (44 %, 52 %, 59 % at 1, 2, and 3 years, respectively) (p < 0.01). At 3 years after implantation, there was no significant difference in mean visual acuity from baseline values in implanted eyes, while mean vision declined in fellow eyes (p < 0.01). Macular edema was evaluated based on the area of hyperfluorescence on FA. Reduction in ME in the implanted eyes was seen in 86 % at 1 year and 73 % at 3 years, versus 28 % at 1 year and 28 % at 3 years in fellow non-implanted eyes. The mean area of ME in implanted eyes maintained a statistically significant decrease lower than baseline at 1, 2, and 3 years after implantation (p < 0.01) [56].

In 2011, the 2-year results from the Multicenter Uveitis Steroid Treatment (MUST) trial were published, including evaluation of ME by OCT. This NIH-sponsored randomized clinical trial compared the safety and efficacy of the FlAc implant to systemic therapy in 255 patients with noninfectious intermediate, posterior, and panuveitis [38]. In this trial, ME was defined as center point macular thickness of greater than 240 microns assessed on Stratus OCT-3. At baseline, the proportion of eyes with ME was similar between the implant group (41 %) and the systemic therapy group (39 %). The implant group showed significantly greater reduction in proportion of eyes with ME compared with the systemic medication group at 6 months (decreased to 20 % and 34 %, respectively, p = 0.002). At 2 years, both groups still showed improvement from baseline (22 % and 30 %), although the difference between the two groups was no longer statistically significant (p = 0.071).