Introduction

figure a

Diabetes is a major disease worldwide, with the International Diabetes Federation reporting that 463 million people suffer from diabetes globally [1]. In Japan, the prevalence of diabetic retinopathy has been estimated to be 10% in all patients with diabetes, with hemoglobin A1c (HbA1c) levels over 6.5% [2]. Diabetic macular edema (DME) is one of the major complications seen in diabetic retinopathy. As DME is a vision-threatening condition, various treatments have been attempted in patients with DME. Recently, anti-vascular endothelial growth factor (anti-VEGF) has become the major treatment for these patients, based on the results of the RISE and RIDE [3], VIVID [4], DRCR-net [5], and Protocol-T studies [6]. Vitrectomies have been performed as a treatment for naïve patients with DME since 1992 [7], because initially the patients were reported to have a thickened and taut premacular posterior hyaloid membrane [7], which was thought to contribute to the development of DME. Subsequently, vitrectomy was used to treat DME without traction on the vitreoretinal interface [8, 9] according to the idea that removal of the vitreous or internal limiting membrane might lead to rapid absorption of DME [10]. In contrast, long-term management of patients with DME has revealed that some become refractory to anti-VEGF agents or periocular steroid therapies. Although these therapies have greatly contributed to improvements in visual function and retinal morphology, there are still cases that continue to have impaired visual outcomes or lingering retinal edema after treatment with anti-VEGF or periocular steroid. Thus, it remains unclear whether vitrectomy is actually beneficial in patients with intractable DME, specifically in patients who were previously treated with anti-VEGF therapy. To clarify this, the present study evaluated and analyzed the surgical outcomes for patients with intractable DME.

Methods

This retrospective study was designed to evaluate the efficacy of vitrectomy in participants with intractable DME. The study protocol was approved by our Institutional Review Board and complied with the ethical standards defined by the Declaration of Helsinki. The study plan was disclosed to all participants prior to their enrollment in the study.

Participants included in the study had to be 20 years of age or older with DME due to diabetes, which was primarily evaluated based on the HbA1c blood concentration. In addition, all participants with DME had to have been treated with anti-VEGF (bevacizumab, ranibizumab, or aflibercept) or periocular triamcinolone acetonide prior to participating in the study. This included participants with DME that failed to resolve even though they had previously undergone these treatments. Furthermore, the participant’s eye had to exhibit subretinal fluid (SRF) and/or intraretinal fluid (IRF) on swept-source optical coherence tomography (SS-OCT; Topcon DRI Triton, Tokyo) imaging. Vitrectomies, including posterior vitreous detachment and internal limiting membrane peeling, were performed by 11 surgeons using the Constellation 25G system (Alcon Laboratories, Inc., USA). Cataract surgery was simultaneously performed in 9 eyes.

The primary measurements analyzed included changes in visual acuity (VA) and retinal morphology. Best-corrected visual acuity (BCVA) of each eye included in study was examined at baseline and at 3, 6, and 12 months postoperatively. Retinal morphological changes were evaluated based on the SS-OCT images at baseline and at 3, 6, and 12 months posttreatment. All cases were followed up for at least 3 months, with a total of 13 eyes (48%) evaluated throughout the entire follow-up period.

One-way analysis of variance was used to analyze whether there was improvement in the VA and a reduction of the retinal thickness after the surgery. Data were analyzed using GraphPad Prism Version 7 software (GraphPad Software, La Jolla, CA, USA). All data are presented as mean ± standard error.

Results

Based on fundus examinations, optical coherence tomography angiography, or fluorescein angiography, 20 eyes were diagnosed with proliferative diabetic retinopathy (PDR), 5 eyes with pre-PDR, and 1 eye with simple diabetic retinopathy.

Prior to surgery, 25 eyes had been treated with anti-VEGF, with a mean of 3.1 ± 2.8 injections. Periocular triamcinolone acetonide was injected into 15 eyes. The average period before entering the study was 21.0 ± 15.0 months (range, 4–61 months). Table 1 summarizes the preoperative demographic characteristics of the patients.

Table 1 Preoperative demographic characteristics: HbA1c Hemoglobin A1c, Cr creatine, VEGF vascular endothelial growth factor, PEA phacoemulsification, IOL intraocular lens. The data was presented as mean ± standard error
Full size table

Best-corrected logMAR visual acuities before and at 3, 6, and 12 months postoperatively were 0.49 ± 0.06, 0.55 ± 0.06, 0.40 ± 0.08, and 0.38 ± 0.08, respectively (Fig. 1). Although there was a tendency for the VA to decrease, there was no significant improvement after surgery. Compared with the preoperative BCVA, the postoperative BCVA improved by more than two lines in 4 eyes (14%), remained the same in 17 eyes (63%), and decreased in 6 eyes (23%) at 3 months postoperatively. Improvements in BCVA at 6 and 12 months were seen in 18% and 23% of cases respectively, with the values increasing during the follow-up (Fig. 2). The central retinal thickness (CRT) values before and at 3, 6, and 12 months postoperatively were 550 ± 24, 472 ± 44, 391 ± 41, and 415 ± 50 μm, respectively. There was a significant decrease (P < 0.05) in the CRT at 6 months compared with that observed at baseline (Fig. 3).

Fig. 1
figure 1

Change in mean best-corrected visual acuity (BCVA) before and at 3, 6, and 12 months postoperatively

Full size image
Fig. 2
figure 2

The proportion of patients with changes in best-corrected visual acuity of 0.2 logMAR vision or more at 3, 6, and 12 months postoperatively

Full size image
Fig. 3
figure 3

Change in central retinal thickness (CRT) before and at 3, 6, and 12 months postoperatively. CRT was significantly reduced at 6 months postoperatively (*P < 0.05)

Full size image

Nine of the 27 eyes had vitreomacular traction (VMT) or epiretinal membrane (ERM) prior to surgery. The best-corrected logMAR improved from 0.56 ± 0.12 at baseline to 0.54 ± 0.10, 0.21 ± 0.09, and 0.14 ± 0.07 at 3, 6, and 12 months postoperatively, with a significant improvement (P < 0.05) found at 12 months after the surgery. In addition, CRT values before and at 3, 6, and 12 months postoperatively were 485 ± 35, 306 ± 41, 361 ± 44, and 331 ± 55 μm, respectively. There was a significant decrease (P < 0.01) in the CRT at 3 months compared to that observed at baseline (Figs. 4 and 5).

Fig. 4
figure 4

The proportion of patients with changes in central retinal thickness (CRT) of 50 μm or more at 3, 6, and 12 months postoperatively

Full size image
Fig. 5
figure 5

Change in mean best-corrected visual acuity (BCVA) and central retinal thickness (CRT) in all cases and in patients with vitreomacular traction (VMT) or epiretinal membrane (ERM) (9/27 eyes) before and at 3, 6, and 12 months postoperatively. A significant improvement in VA at 12 months postoperatively (P < 0.05), and a significant decrease in CRT at 3 months postoperatively (P < 0.01) were seen in patients with VMT or ERM

Full size image

Discussion

The results of this retrospective study suggest that vitrectomy is effective in patients with intractable DME. The retinal anatomy in 78% of the patients had stabilized by 12 months postoperatively. However, there was no significant improvement in BCVA after the surgery.

Since the 1990s, vitrectomy has been performed in patients with DME based on the observation that the attached posterior vitreous might be highly associated with the formation of DME [7]. Similarly, Sebag and Balazs reported finding cases in which the remaining vitreous cortex on the macula resulted in persistent macular edema [11]. Subsequently, Yamaguchi et al. reported cases in which the vitreofoveal separation potentially led to resolution of the DME [12]. Based on these reports, the beneficial effect of using vitrectomy for DME was widely evaluated. Otani and Kishi reported on the visual outcomes of vitrectomy for treatment naïve DME. In their study, patients were followed up for 5 months after the surgery. Although there was no significant improvement in the mean VA after surgery, improvement in VA of more than two lines was achieved in 57% of the patients. Anatomically, they reported that the CRT decreased from 617 to 546 μm [13]. Heij et al. followed up 19 patients with DME for a median of 11 months postvitrectomy and demonstrated that there was significant improvement in BCVA, with 71% of the eyes showing an improvement in VA of one line or more. Additionally, the macular edema resolved in 95% of the patients at the final visit [14]. Notably, the patients included in this study did not have evident VMT. It has been reported in previous studies that patients attained anatomical improvement after undergoing vitrectomy [14]. In contrast, the postoperative BCVA and the frequencies of the improvement in BCVA were limited in the present study. This suggested that chronic DME, as seen in patients in the present study, might affect preoperative macular function, even though there were some patients that sustained a moderate BCVA from beginning of the study. This speculation is based on the pathogenesis of DME, in which multiple factors such as ischemia of the retina, accumulation of hard exudate, and other general conditions have been shown to affect macular function [15].

In 9 of 27 eyes with VMT or ERM that were examined in our study, BCVA was significantly improved at 12 months postoperatively. This result is consistent with the findings reported by a previous study [16]. Moreover, our results showed that the good VA was maintained even at 12 months. It should be noted that in cases with DME associated with VMT or ERM, we observed recurrence of DME postoperatively during the 12 months of follow-up. It was necessary to treat these patients with additional anti-VEGF or periocular corticosteroid. However, the average number of additional postoperative treatments between patients with VMT or ERM and those without tractions was 0.56 ± 0.73 and 2.44 ± 2.85, respectively, even though the follow-up period after the surgery differed between the two groups (7.3 ± 4.5 months versus 8.0 ± 4.2 months, respectively).

The pharmacokinetic properties of anti-VEGF [17] likely mean it has a shorter half-life in the vitrectomized eyes. Consequently, administration of periocular steroid will potentially be a more beneficial strategy when taking into consideration the longer management that patients with DME must undergo.

Based on the findings of previous reports, we removed the internal limiting membranes (ILMs) in all cases in the present study. Gandorfer et al. were the first to demonstrate the efficacy of ILM removal, with 11 of 12 eyes showing improvement in VA by at least two lines during the follow-up period (mean 4 months) [17]. Subsequently, several other institutions have reported similar good results [18,19,20,21]. One of the beneficial advantages of removing the ILM is that it not only can prevent the postoperative formation of the epimacular membrane [17] but it also can potentially lead to a complete release of the tractional force from the vitreous and rapid resolution of diffuse macular edema [7]. Recently, Imai et al. reported that en bloc removal of a cystoid lesion combined with vitrectomy can lead to a sustained VA and a stabilized retinal thickness postoperatively [22]. In addition, these authors showed that the encapsulated fibrinogen consisted of a cystoid lesion. In the present study, several patients exhibited a recurrence of DME after the initial surgery, so this new technique might be effective in these types of patients.

The limitations of the present study included its retrospective design and the inclusion of only a small number of subjects in a single-arm modality of treatment. In the future, a prospective comparative study will need to be conducted to definitively clarify the present findings.

In conclusion, vitrectomy for intractable DME can sustain and stabilize both visual function and retinal morphology, and may be specifically beneficial for DME patients with VMT and ERM.