Early postoperative retinal thickness changes and complications after vitrectomy for diabetic macular edema

Methods 

The study population consisted of 65 eyes of 63 consecutive cases in which a pars plana vitrectomy was performed for a decrease in diffuse diabetic macular edema. Vitrectomy was performed at Toho University Sakura Hospital and at Yamagata University between October 1998 and June 2001. The follow-up interval ranged from 6 to 36 months (mean ± standard deviation [SD] follow-up, 12.6 ± 7.4 months).

Twenty patients were women and 43 were men, and their ages ranged from 31 to 80 years (60.4 ± 8.9 years [mean ± SD]). One eye had moderate nonproliferative diabetic retinopathy (NPDR), severe NPDR was observed in 34 eyes, and proliferative diabetic retinopathy was observed in 30 eyes. Panretinal photocoagulation had been performed in 64 eyes, direct photocoagulation to microaneurysms in the macula had been performed in 48 eyes, and focal/grid photocoagulation had been performed in 5 eyes before entry into this study. The cataract extraction with implantation of a posterior chamber intraocular lens had been performed previously in 18 of 65 (28%) eyes. The complete posterior vitreous detachment with a Weiss ring had been observed in 24 eyes.

All patients were evaluated preoperatively and postoperatively by stereoscopic biomicroscopy, color fundus photography of the macula, and by OCT to determine the foveal configuration and retinal thickness. The retinal thickness was defined as the distance between the vitreoretinal interface and the retinal pigment epithelium in the center of the fovea. The best-corrected visual acuity was examined using the decimal visual acuity system in 29 eyes, which was converted to the logarithm of the minimum angle of resolution (logMAR) scale. In 36 eyes, Early Treatment of Diabetic Retinopathy Study visual acuity charts were used, and visual acuities were shown in logarithmic scale. In all the patients, the best-corrected visual acuity was measured by examiners unrelated to the coauthors.

The inclusion criterion for the subjects was diffuse diabetic macular edema with the diffusion caused by hyperpermeability of the entire dilated perimacular capillary bed due to the breakdown of the inner blood–retina barrier. The indications for pars plana vitrectomy in this study were diffuse diabetic macular edema, preoperative visual acuity less than 20/40, and noneffective macular photocoagulation therapy.

The surgical technique was a three-port pars plana vitrectomy. The cataract extraction during surgery had been performed in 28 of 65 (43%) eyes. In 19 eyes (29%) without cataract before surgery, the clear lens was retained in the course of the follow-up period. In all eyes, the posterior hyaloid membrane was separated from the retinal surface with a microvitreoretinal blade or with suction from the vitreous cutter. In the present study, the intentional removal of the ILM was not performed in any cases. However, we could not deny an accidental removal of the ILM during surgery. In 20 eyes with an epimacular membrane, the membrane was removed intraoperatively and prepared for histopathological observations.

Results 

The mean best-corrected visual acuity at the most recent follow-up examination was 0.696 ± 0.491 (logMAR units; mean ± SD), which was significantly better than the preoperative mean best-corrected visual acuity of 0.827 ± 0.361 (P < .0001; Wilcoxon signed-ranks test). Preoperative best-corrected visual acuity and postoperative best-corrected visual acuity are shown in Figure 1. The final visual acuity improved by 2 or more lines in 32 of 65 eyes (45%), remained unchanged in 32 of 65 eyes (49%), and was exacerbated after the surgery in 4 of 65 eyes (6%) owing to neovascular glaucoma (2 eyes) and residual cystoid macular edema (2 eyes).

View Large Image Download to PowerPoint

FIGURE 1. Scattergram of visual acuity before and after surgery (visual acuity is expressed in logarithmic units).

The preoperative mean foveal retinal thickness of 463.7 ± 177.3 μm was significantly thicker than the postoperative mean foveal thickness of 224.9 ± 116.9 μm (P < .0001; Wilcoxon signed-rank test). In this study, the changes in foveal retinal thickness in the early stage were investigated in 30 eyes, which were in good condition for OCT examination. The changes in the retinal thickness were followed from the preoperative examination (baseline level) to that at 1 week, and at 1, 2, 3, and 4 months postoperatively. These findings showed that the mean postoperative retinal thickness had decreased significantly 7 days after surgery, then remained unchanged for approximately 1 month, and thereafter gradually decreased until 4 months (FIGURE 2, FIGURE 3).

View Large Image Download to PowerPoint

FIGURE 2. (A) Preoperative and (B) postoperative optical coherence tomography images of the retina of the right eye of a 70-year-old man. The postoperative retinal thickness is thinner than the preoperative retinal thickness at 7 days after surgery. Thereafter, the retinal thickness remains unchanged for 1 month after surgery (C), and then decreases gradually until 4 months (D). Best-corrected visual acuity (logMAR) is 0.5 at preoperative examination, 0.8 at 6 days after surgery, 0.5 at 1 month after surgery, 0.7 at 3 months after surgery, and 0.1 at 9 months after surgery respectively.

View Large Image Download to PowerPoint

FIGURE 3. The preoperative and postoperative mean retinal thickness changes of 30 eyes. *P = .0263; **P = .0464; determined by Wilcoxon signed-rank test.

Intraoperative and postoperative complications were observed in 21 eyes of 21 cases. A peripheral retinal tear was detected intraoperatively in 3 of 65 (4.6%) eyes, a postoperative rhegmatogenous retinal detachment in 1 of 65 (1.5%) eyes, a postoperative epiretinal membrane in 9 of 65 (13.8%) eyes, neovascular complications in 3 of 65 (4.6%) eyes, recurrent vitreous hemorrhage in 1 of 65 (1.5%) eyes, hard exudate deposits in the center of the macula in 3 of 65 (4.6%) eyes, and lamellar macular hole in 1 of 65 (1.5%) eyes. The intraoperative retinal break was repaired successfully with the removal of vitreous around the retinal tear, endolaser retinal photocoagulation, and fluid-air exchange.

Neovascularization of the iris was not observed in any eye preoperatively. However, rubeosis iridis was observed in three eyes of three cases within 1 to 2 months after the surgery. Preoperatively, these three patients had proliferative diabetic retinopathy and had had panretinal photocoagulation with 1,500 to 2,000 spots. Two of the three patients had undergone hemodialysis. In one, carotid arteriography showed 65% stenosis of the right internal carotid artery (right eye was the affected eye) and the arm-to-retina time was prolonged to 24.2 seconds in fluorescein angiography. Retinal photocoagulation was performed as soon as the iris neovascularization was detected at the postoperative examination, and the iris neovascularization regressed from 4 to 8 weeks after the photocoagulation in all eyes. Intraocular pressure was kept below 20 mm Hg with no medication in one eye; however, the other eye required eyedrops to control the intraocular pressure.

Postoperative hard exudate deposits in the center of the macula were observed in three eyes of three cases, and many small hard exudates in the macular area preceded the development of the hard exudates. Serous retinal detachment was not observed in the macular area preoperatively. Although the hard exudates cleared 3 months after surgery, chorioretinal atrophy was observed at the site of the lesions in one eye.

Postoperative epiretinal membranes were observed in nine eyes. Five membranes were detected in eyes with and four membranes in eyes without posterior vitreous detachment preoperatively. The residual posterior vitreous cortex was found on the surface of the retina intraoperatively in two eyes with a posterior vitreous detachment and two eyes without posterior vitreous detachment. In the other five eyes with a postoperative epiretinal membrane, the retinal surface was not examined carefully intraoperatively.

In an illustrative case, a 64-year-old man had a diffuse diabetic macular edema with large cystoid spaces in his left eye (Figure 4, top). A posterior vitreous detachment and an epiretinal membrane were observed, and his left best-corrected visual acuity was 20/200. A standard three-port pars plana vitrectomy with removal of the epimacular membrane was performed. A macular hole was not detected preoperatively or intraoperatively. One week after the surgery, the OCT foveal images appeared similar to those with a lamellar macular hole, such as the disappearance of the inner retinal layer and the existence of the outer layer of retina (Figure 4, middle). At this time, the best-corrected visual acuity was not changed at 20/200. At 1 month after surgery, the OCT images had the characteristics of a lamellar macular hole with a foveolar-like slope (Figure 4, bottom). The postoperative best-corrected visual acuity was maintained at 20/200 at 14 months after vitrectomy.

View Large Image Download to PowerPoint

FIGURE 4. Lamellar macular hole formation. (Top) Optical coherence tomography shows diffuse macular edema with large cystoid space and epiretinal membrane(white arrows). (Middle) Optical coherence tomography shows a lamellar macular hole. (Bottom) Optical coherence tomography shows a lamellar macular hole with a foveolar-like slope.

Light microscopic examination showed that the epiretinal membrane obtained from this patient was homogeneous with an ILM-like structure with cells arranged in a single layer. Several parts of this membrane stained positive with periodic acid-Schiff ([PAS] Figure 5).

View Large Image Download to PowerPoint

FIGURE 5. Photomicrograph of an epiretinal membrane from a 64-year-old man. Tissue is homogeneous and has an internal limiting membrane-like structure with cells arranged in a single layer. Several parts of the membrane stained positive with periodic acid-Schiff (arrows; ×150).

Discussion 

Our results have demonstrated that the retinal thickness is reduced within 7 days postoperatively with OCT examination. The reason why the retinal thickness decreased so quickly after vitrectomy is unknown but we suggest that the high concentrations of oxygen supplied from the ciliary body causes retinal vasoconstriction, resulting in retinal thickness decreases in the early period after vitrectomy.6, 7, 8 Afterward, the retinal thickness kept on decreasing for at least 4 months. These findings mean that the evaluation of the effectiveness of vitrectomy for diabetic macular edema by retinal thickness changes should be performed no sooner than 4 months after vitrectomy.

Several intraoperative and postoperative complications have been reported after pars plana vitrectomy for diabetic macular edema. These complications included intraoperative retinal tears,3 postoperative rhegmatogenous retinal detachment,1, 3, 4 neovascular glaucoma,3 postoperative vitreous hemorrhage,1, 2, 3, 4 postoperative epiretinal membrane formation,1, 2, 3 cataracts,1, 3, 4 and macular ischemia.2

Rubeosis iridis was observed in 3 of 65(4.6%) eyes postoperatively. Previously, it was reported that the incidence of postoperative rubeosis of the iris and neovascular glaucoma after vitrectomy was 5% to 22%.9, 10, 11 The appearance of rubeosis iridis was within 1 to 2 months after the surgery in our three cases indicating that frequent anterior segment examination should be performed during the first few months after the operation. A carotid artery obstruction was detected in one case. Because carotid artery obstruction causes neovascular glaucoma,12, 13, 14 the cases with an increased arm-to-retina time in fluorescein angiography should be checked for carotid artery obstruction before the surgery. If there are extensive ischemic changes, full panretinal photocoagulation should be performed preoperatively and intraoperatively.

We observed postoperative hard exudate deposits in the center of the macula in 3 of 65 (4.6%) eyes. Otani and associates15 reported that subretinal hard exudates accumulated after absorption of the subretinal serous fluid. However, in our three eyes, serous retinal detachment was not observed in the preoperative and postoperative examinations.

Postoperative epiretinal membranes were observed in nine eyes with or without posterior vitreous detachment. We noticed residual vitreous cortex on the surface of the retina in several eyes with or without posterior vitreous detachment intraoperatively. Intraocular inflammation, retinal photocoagulation, vascular disease, and retinal breaks can cause glial proliferation on the retinal surface.16, 17, 18 We suggest that epiretinal membranes are formed by cellular proliferation due to surgical inflammation, and the cells scaffold to the residual vitreous cortex onto the retina surface.

A lamellar macular hole associated with vitrectomy was observed in one eye. A lamellar macular hole due to vitrectomy for diabetic macular edema has never been reported (from a MEDLINE search) and should be added to the list of complications of vitrectomy. In this study, vitrectomy with removal of an epiretinal membrane may have caused the lamellar macular hole in a case with large cystoid spaces composed of thin inner retinal layer preoperatively. The formation of a macular hole after epiretinal membrane removal in nondiabetic eyes19 but with major cystoid macular edema20 was reported. In our cases, it is possible that the tissue removed may have contained the internal limiting membrane because a PAS-positive membrane-like structure21, 22 was detected in the histopathologic examination. These findings suggest that the peeling of epiretinal membranes and ILMs must be done carefully especially in eyes with cystoid macular edema with a thin inner retinal wall.

In conclusion, these results indicate that careful examination is needed to determine whether pars plana vitrectomy is efficacious for diabetic macular edema because there are severe complications. Furthermore, careful observations of the changes in the retinal thickness in the postoperative period may help in determining how vitrectomy improves diabetic macular edema.