Retinal architecture as a determinant of brilliant blue G phototoxicity during internal limiting membrane peeling: A case series

Abstract

BACKGROUND

Chromovitrectomy using vital dyes has become integral to internal limiting membrane (ILM) peeling in macular surgery. Although brilliant blue G (BBG) is considered safer than earlier dyes, emerging clinical and experimental evidence suggests that dye-light interactions during surgery may result in retinal phototoxicity under specific conditions.

AIM

To describe the retinal phototoxicity features associated with BBG dye in patients undergoing ILM peeling for macular surgeries.

METHODS

This observational, comparative, retrospective case series analyzed 9 eyes of 9 patients demonstrating retinal phototoxicity after uneventful BBG-assisted ILM peeling. Surgical indications included full-thickness macular hole (MH) in 4 eyes, epiretinal membrane associated retinal traction in 4 eyes, and vitreomacular traction in 1 eye. Retinal phototoxicity was assessed during clinical examination, using optical coherence tomography, and fundus autofluorescence imaging. Parameters such as preoperative and postoperative visual acuity (VA), macular thickness, BBG exposure characteristics (single/repeated staining, exposure duration, staining medium), and follow-up outcomes were recorded. Cases were grouped based on surgical indications: Group A (MH, 4 eyes) and group B (non-MH, 5 eyes).

RESULTS

In group A, all cases showed foveal phototoxic damage along with macular involvement despite successful MH closure. Two eyes experienced worsened VA, while the other two maintained baseline VA. In group B, phototoxicity involved the macular region but spared the fovea; VA improved in 3 eyes (60%) and was maintained in 2 eyes (40%). Retinal toxicity, characterized by photoreceptor and retinal pigment epithelium damage, was evident within 1-4 weeks postoperatively.

CONCLUSION

BBG-related retinal phototoxicity may occur following ILM peeling, with eyes undergoing MH surgery appearing more vulnerable, possibly due to the absence of protective foveal neurosensory tissue. Awareness of this potential risk may encourage surgeons to consider minimizing illumination intensity and dye exposure, and to adopt protective intraoperative strategies aimed at preserving foveal integrity and optimizing visual outcomes.

Key Words: Retinal toxicity; Brilliant blue G; Endo illumination; Outcome; Fovea

Core Tip: This case series identifies retinal architecture as a key determinant of brilliant blue G (BBG)-related toxicity during internal limiting membrane peeling. Phototoxicity was more severe in macular hole (MH) cases where the exposed retinal pigment epithelium allowed direct BBG-light interaction, leading to foveal damage and poor visual recovery. Non-MH eyes, including epiretinal membrane and vitreomacular traction, retained foveal protection and demonstrated better postoperative vision. The study reinforces careful control of illumination and BBG exposure in MH surgery.

INTRODUCTION

The primary objective of surgery for vitreomacular interface disorders is to relieve anteroposterior traction through pars plana vitrectomy with posterior vitreous detachment induction, and tangential traction through epiretinal membrane (ERM) and internal limiting membrane (ILM) peeling, thereby restoring normal foveal architecture and improving visual acuity[1,2]. ILM peeling is now a widely established technique in the management of macular holes (MHs), vitreomacular traction (VMT), macular puckers, ERMs, tractional and selected non-tractional diabetic macular edema, myopic tractional maculopathy (MTM), optic disc pit maculopathy (ODPM), and sub-ILM hemorrhages such as those seen in Valsalva retinopathy and Terson’s syndrome[3].

Vital dyes including indocyanine green (ICG), brilliant blue G (BBG), and bromophenol blue have been routinely used to facilitate ILM visualization during surgery[4-6]. However, toxic effects on photoreceptors and the retinal pigment epithelium (RPE) have been documented, particularly with ICG, with both experimental and clinical evidence demonstrating retinal damage and postoperative visual dysfunction[5,7,8]. Consequently, ICG use is no longer recommended for macular procedures. BBG has demonstrated a comparatively favourable safety profile with superior anatomical and functional outcomes relative to ICG[4,9,10], although reports of retinal toxicity do exist. Studies described inner retinal thinning following BBG use in MH surgery, and our group has previously reported both acute and long-term phototoxic effects of BBG in MH and ERM surgeries[11-14].

Although the surgical steps involved in managing vitreomacular interface disorders are largely similar, the underlying foveal anatomy varies substantially across disease entities. Eyes with full-thickness or lamellar MHs are characterized by focal loss of neurosensory foveal tissue, whereas conditions such as ERM, VMT, MTM, and ODPM typically retain increased foveal thickness. Since BBG-associated phototoxicity predominantly affects the photoreceptors and RPE, differences in retinal architecture may critically influence tissue vulnerability to dye- and light-mediated injury. While existing literature has focused mainly on dye concentration, exposure duration, and endoillumination parameters as determinants of toxicity, the impact of foveal structural integrity has not been systematically examined. Evaluating BBG-related phototoxicity in eyes with and without MHs therefore provides a unique opportunity to assess the role of retinal architecture in modulating susceptibility to injury. Such an analysis is clinically relevant, as it may identify anatomical risk factors that inform surgical planning and guide intraoperative strategies aimed at minimizing retinal damage.

To explore this, we conducted a retrospective analysis of BBG-associated retinal phototoxicity in eyes undergoing ILM peeling across various macular surgery indications.

MATERIALS AND METHODS

This retrospective case series included patients diagnosed with BBG-related retinal phototoxicity following ILM peeling for various macular pathologies between 2017 and 2022. Phototoxicity was defined as outer retinal and/or RPE damage detected during postoperative follow-up. Diagnosis was based on clinical observation of pigmentary changes corresponding to the ILM-peeled area and confirmed using multimodal imaging, including mottled hyper- and hypo-autofluorescence on fundus autofluorescence (FAF) and loss of outer retinal layers with irregular RPE alterations on optical coherence tomography (OCT). The study adhered to the tenets of the Declaration of Helsinki. Institutional ethics approval was obtained (EC Ref. No: C/2019/03/04) from the Narayana Nethralaya Ethics Committee, Bangalore. Owing to the retrospective design, informed consent was waived.

Following identification of eligible cases, clinical records were reviewed to extract demographic details, preoperative clinical and imaging characteristics, intraoperative parameters, and postoperative outcomes. Demographics included age and sex. Preoperative variables included eye laterality, best-corrected visual acuity, surgical indication, OCT assessment of the outer retina and RPE, and central foveal thickness. Central macular thickness was measured from the ILM to the inner RPE boundary using built-in callipers on the OCT device.

All patients underwent standard 25-gauge three-port pars plana vitrectomy using the Alcon CONSTELLATION^® Vision System (TX, United States). Posterior vitreous detachment was induced in all cases using intravitreal triamcinolone acetonide. ILM peeling was performed after staining with BBG. Intraoperative details recorded included whether BBG (Ocublue plus 0.05% w/v, Aurolab, Madurai, India) staining was single or repeated, staining under air or balanced salt solution (BSS), duration of intraocular dye exposure, total surgical time, and endotamponade type. All procedures were performed by a single experienced surgeon.

Postoperative assessment was limited to clinical examination, optical coherence tomography, and FAF imaging, as additional functional and microvascular investigations were not routinely performed during the study period. Postoperative data included anatomical resolution of the primary pathology, characteristics and severity of BBG-related phototoxicity including foveal involvement (defined as outer retinal loss, RPE thinning, and increased choroidal hyper transmission on OCT), time to recognition of toxicity, and best-corrected visual acuity at early postoperative review (4-6 weeks) and final follow-up. Total follow-up duration was documented.

To evaluate whether anatomical substrate influenced phototoxic patterns and visual outcomes, cases were categorized into two groups based on indication: (A) MH; and (B) non-MH. Given the retrospective design of the study and the rarity of clinically evident BBG-related phototoxicity, the inclusion of a matched control group was not feasible.

Statistical analysis

Data analysis was conducted using XLSTAT software (Lumivero, Denver, CO, United States version 2023.3.1.1416). Data distribution was summarized as medians with interquartile ranges for continuous variables, while categorical variables were presented as n (%).

RESULTS

Demographic data

The study identified 9 eyes (9 cases) with outer retinal layer and RPE damage secondary to BBG-related retinal phototoxicity. The median age of the participants was 69.0 years, ranging from 52 years to 86 years. Table 1 provides a complete description of the cases included in the study. Based on the categorization, 4 eyes were assigned to group A and the remaining 5 eyes to group B. Group B included 4 (80%) cases with idiopathic ERM and 1 (20%) case of VMT.

Table 1 Information of the cases included in the study.

Case No.	Age	Sex	Eye	Reason for ILM peeling	Study group	Preop VA	Pre-op CMT	Pre-op ORL status	Pre-op RPE layer healthy	BBG exposure time (minute)	BBG stain under	Stain	Tamponade agent	Total surgery time (minute)	Time gap post-surgery after which retinal damage was noted (weeks)	Retinal toxicity involving the fovea	Surgical result	Early post op VA	Last visit VA	F/U time (months)
1	69	F	RE	Idiopathic MH	A	6/36	0	CBA	Yes	2	AIR	SINGLE	SF6	90	1	Yes	Closed MH	6/36	6/36	60
2	75	M	RE	Idiopathic MH	A	6/24	0	CBA	Yes	2	AIR	SINGLE	SF6	75	3	Yes	Closed MH	6/36	6/36	9
3	77	M	LE	Idiopathic MH	A	6/120	0	CBA	Yes	2	AIR	SINGLE	SF6	45	2	Yes	Closed MH	6/120	6/120	60
4	66	F	LE	Idiopathic MH	A	6/15	0	CBA	Yes	2	BSS	SINGLE	C3f8	65	4	Yes	Closed MH	6/24	6/24	8
5	86	F	LE	ERM with increased retinal traction	B	6/120	515	N	Yes	2	AIR	DOUBLE	SF6	90	1	No	Retinal thickness reduced	6/60	6/12	14
6	77	M	RE	ERM with increased retinal traction	B	6/60	518	N	Yes	2	AIR	DOUBLE	SF6	60	1	No	Retinal thickness reduced	6/60	6/60	48
7	52	F	RE	ERM with increased retinal traction	B	6/30	957	N	Yes	2	AIR	DOUBLE	AIR	80	4	No	Retinal thickness reduced	6/30	6/30	22
8	67	F	LE	ERM with increased retinal traction	B	6/30	590	N	Yes	3.5	AIR	DOUBLE	BSS	110	1	No	Retinal thickness reduced	6/24	6/8	10
9	65	F	LE	Vitreofoveal traction with increased retina thickness	B	6/18	468	N	Yes	2	BSS	SINGLE	C3f8	65	4	No	Normal foveal contour	6/9	6/9	5

M: Male; F: Female; RE: Right eye; LE: Left eye; ILM: Internal limiting membrane; VA: Visual acuity; CMT: Central macular thickness; ORL: Outer retinal layers; RPE: Retinal pigment epithelium; BBG: Brilliant blue G; F/U: Follow-up; MH: Macular hole; ERM: Epiretinal membrane; CBA: Cannot be assessed; N: Normal; BSS: Balanced salt solution; SF6: Sulfur hexafluoride; C3F8: Perfluoroproprane.

MH cases (group A)

Group A included four eyes (44%). Preoperative best-corrected visual acuity ranged from 20/50 to 20/400. Preoperative assessment of central macular thickness and outer retinal integrity was not feasible due to the presence of a MH; however, RPE reflectivity appeared normal on OCT in all eyes. Intraoperatively, ILM staining with BBG was performed once in all cases under air in three eyes (75%) and under BSS in one eye (25%). The duration of dye exposure was consistently 2 minutes. Endotamponade selection varied: Sulfur hexafluoride (SF₆) was used in three eyes (75%) and perfluoropropane (C₃F₈) in one eye (25%). Median surgical time was 70 minutes (range: 45-90 minutes). Postoperatively, all MHs achieved successful closure with restoration of foveal architecture. Signs of BBG-related RPE toxicity were detected as early as 1 week after surgery, following gas absorption, based on clinical examination, OCT, and FAF findings. Two eyes (50%) demonstrated a decline in visual acuity, while the remaining two eyes (50%) showed no improvement at early or final follow-up. Median follow-up duration in this group was 34.5 months (range: 8-60 months).

Non-MH cases (group B)

Group B comprised five eyes (56%). Preoperative best-corrected visual acuity ranged from 20/60 to 20/400, and central macular thickness measured 468-957 µm. The outer retinal layers and RPE reflectivity were intact on preoperative OCT in all cases. ILM staining with BBG was performed under air in four eyes (80%) and under BSS in one eye (20%). Staining was performed twice in four eyes (80%) and once in one eye (20%). The duration of dye exposure was 2 minutes in four cases (80%) and 3.5 minutes in one case (20%). Endotamponade selection included SF₆ in two eyes (40%) and C₃F₈, BSS, and air in one eye each (20%). The median surgical duration was 80 minutes (range: 60-110 minutes). All cases demonstrated favourable postoperative anatomical outcomes, including reduction in retinal thickness and successful retinal reattachment. BBG-related RPE toxicity was identified as early as 1 week postoperatively in all cases based on multimodal imaging. Visual acuity improved in three eyes (60%) and remained stable in two eyes (40%). Notably, RPE and photoreceptor damage spared the foveal center in all eyes. Median follow-up duration for this group was 14 months (range: 5-48 months). Figures 1, 2, and 3 provides the images of the case examples included in the study.

Figure 1 Imaging features before and after surgery - case 2. This figure corresponds to patient 2, who underwent surgery for an idiopathic macular hole (MH) in the right eye with an initial visual acuity of 6/24. A: Preoperative optical coherence tomography (OCT) revealed a fully detached posterior vitreous cortex, intraretinal cystic spaces, and intact retinal pigment epithelium (RPE) at the fovea. Internal limiting membrane (ILM) staining was performed under air for 2 minutes. The total surgical time was 95 minutes; B and C: One month postoperatively, OCT showed a closed MH with photoreceptor and RPE damage in the foveal and parafoveal regions, corresponding to the ILM peel area, with visual acuity of 6/36; D-F: At 3- and 9-months post-surgery, the hole remained closed, but progressive outer retinal and RPE thinning with increased choroidal hyperreflectivity was observed. Visual acuity at the final visit remained 6/36.

Figure 2 Imaging features before and after surgery - case 8. A and B: Correspond to patient 8, who underwent surgery for an epiretinal membrane in the left eye, presenting with retinal traction and increased retinal thickness (590 µm). The initial visual acuity was 6/30. Preoperative optical coherence tomography (OCT) showed significant retinal traction caused by the thick epiretinal membrane, with intact outer retinal layers and retinal pigment epithelium (RPE) at the fovea. ILM staining was performed twice under air for a total of 3.5 minutes, with a surgical time of 110 minutes; C and D: Ten days postoperatively, OCT revealed outer retinal layer damage and RPE mottling, sparing the fovea. The parafoveal damage corresponded to the ILM peel area, with visual acuity improving to 6/24; E and F: At 10 months post-surgery, further reduction in foveal thickness was observed, with visual acuity improving to 6/8 and the brilliant blue G related retinal toxicity sparing the fovea.

Figure 3 Imaging features before and after surgery – case 9. A-C: Preoperative imaging of patient 9, who underwent surgery for vitreomacular traction (VMT) with foveal detachment in the left eye. Initial retinal thickness measured 468 µm, and visual acuity was 6/18. Preoperative optical coherence tomography (OCT) shows significant VMT with neurosensory detachment involving the fovea, intact outer retinal layers, intact retinal pigment epithelium (RPE), and a disrupted external limiting membrane. Fundus autofluorescence imaging of the left eye normal autofluorescence signal patterns at the fovea. Internal limiting membrane (ILM) staining was performed under balanced salt solution for 2 minutes, with a total surgical time of 2 hours. One-month postoperative OCT showing brilliant blue G (BBG)-related retinal toxicity sparing the fovea; D-F: At 5 months post-surgery, the foveal contour was restored with visual acuity improvement to 6/9. Fundus autofluorescence identified abnormal hyper- and hypoautofluorescent signals indicative of BBG-related retinal toxicity. OCT at 5 months demonstrated outer retinal damage and RPE mottling, sparing the fovea. The parafoveal damage corresponded to the ILM peel area.

DISCUSSION

This study identified nine cases of BBG-related retinal phototoxicity following vitreoretinal surgery for macular disease. Among eyes undergoing MH repair, phototoxic damage involved the fovea in addition to surrounding macular areas, whereas foveal involvement was not observed in non-MH cases. This pattern is unexpected, as the fovea representing the thinnest retinal region with taller and densely pigmented RPE cells would typically be anticipated to demonstrate greater susceptibility to phototoxic injury.

Previous research has examined the relationship between BBG concentration, exposure duration, and retinal toxicity. In an in vitro study, Penha et al[15] demonstrated that BBG concentrations below 0.25 mg/mL were non-toxic to RPE cells and may even confer a protective effect through upregulation of the anti-apoptotic protein Bcl-2[15,16]. These findings suggest that BBG alone, at concentrations routinely used in macular surgery, is unlikely to induce retinal toxicity. Also, our study findings demonstrated that BBG-related retinal toxicity was not influenced by specific ILM staining techniques and indicated that repeated or prolonged BBG staining, regardless of the medium, was not solely responsible for retinal damage.

The main contributors to BBG-induced retinal toxicity appear to be the characteristics of endoillumination, particularly the intensity and duration of light exposure, as well as the proximity of the endoilluminator to the retinal surface during surgery[11,12,17]. In all cases included in the current series, the total surgical duration was prolonged, ranging from 45 minutes to 110 minutes. The cumulative photoreceptor and RPE injury observed may be attributable to prolonged focal high-intensity endoillumination during extended surgical time. The phototoxic effects of BBG can be attributed to its light absorption properties and emission spectrum, which ranges from 260 nm to 900 nm depending on the solvent used. All BBG solutions exhibit a double-peak absorption curve, with the first peak observed between 260 nm and 280 nm, and the second peak occurring between 540 nm and 680 nm[8]. Endoilluminator emitting light with wavelengths ≥ 400 nm is known to induce phototoxic effects, particularly in the RPE. The CONSTELLATION^® Vision System from Alcon (TX, United States), which utilizes xenon light for endoillumination during vitreoretinal surgeries, has a peak wavelength of 450 nm, with a broader range from 420 nm to 700 nm[16]. Xenon light’s interaction with BBG plays a critical role in phototoxicity. The increased absorption of xenon light by BBG, coupled with alterations in BBG’s emission spectrum, leads to the production of these reactive oxygen species and toxic free radicals, which in turn cause damage to RPE cells and photoreceptors. In vitro studies using human RPE cells (ARPE-19) have demonstrated that BBG exposure significantly reduces RPE cell viability, particularly when focal high-intensity illumination is applied for more than 5 minutes, or medium diffuse illumination is used for over 15 minutes. Long-term damage to photoreceptors and the RPE leads to progressive thinning of the choroidal layers. This occurs as the compromised RPE and photoreceptors demand less nutritional support from the choroid, resulting in subsequent choroidal atrophy[13]. These findings highlight the importance of understanding the interaction between BBG and endoilluminator light in minimizing phototoxic risk during vitreoretinal surgery[8,18].

In this study, fovea-involving toxicity occurred exclusively in MH cases, whereas BBG-related toxic foveal damage was not observed in cases of ERM, MTM, ODPM, or MH-associated retinal detachment, despite routine ILM peeling with BBG in these conditions. This pattern suggests that greater retinal thickness may confer a protective effect against BBG-induced phototoxic injury, contributing to more positive visual outcomes. This observation may be explained by two mechanisms: (1) ERMs demonstrate poor affinity for BBG staining, reducing the amount of dye adhering to the foveal surface; and (2) Increased retinal thickness acts as a physical buffer, shielding the RPE from BBG-enhanced phototoxic stress. Conversely, in MH cases, the absence of foveal tissue leaves the BBG-coated RPE directly exposed to endoillumination, reactive oxygen species, and free radical formation, resulting in greater foveal damage and poorer functional recovery. Collectively, these findings highlight the protective role of retinal thickness and structural architecture in modulating foveal susceptibility to BBG-induced phototoxicity during vitreoretinal surgery.

Although BBG-related injury results in immediate cellular damage to photoreceptors and the RPE with corresponding functional impact, structural changes become clinically apparent only later and may be influenced by the type of endotamponade used[15,19]. In most macular procedures involving BBG-assisted ILM peeling, gas endotamponade is employed, which obscures early postoperative visualization and delays recognition of dye-related toxicity[20]. Structural alterations are more readily detected on OCT and FAF when the damage is parafoveal. In cases with fovea-involving toxicity, OCT proved more sensitive than FAF in detecting early structural disruption. This reduced FAF sensitivity likely reflects difficulty distinguishing true hypoautofluoroscence from damaged RPE vs the physiologic hypoautofluoroscence produced by the densely pigmented foveal RPE.

Additional assessment modalities such as OCT angiography, microperimetry, chromatic perimetry, and focal electroretinography may have provided further insights into microvascular alterations and structure-function correlation in eyes with BBG-related toxicity. However, these investigations were not included in the retrospective review protocol. Future prospective studies incorporating these modalities may better delineate the functional consequences of dye- and light-induced retinal injury.

This study is limited by its retrospective design, small sample size, and absence of a control group. Functional assessments such as microperimetry and electrophysiology were not available, and the use of gas endotamponade may have delayed early postoperative recognition of BBG-related retinal toxicity. Despite these limitations, the study offers important strengths. The consistent multimodal imaging findings across cases allow meaningful pattern recognition, enabling direct comparison of BBG-related phototoxicity in eyes undergoing ILM peeling for MH vs non-MH indications. Our observations suggest that BBG-induced retinal phototoxicity is not solely dependent on dye exposure or endoillumination parameters but is also influenced by underlying retinal architecture. In particular, the increased vulnerability of MH eyes to foveal injury, likely due to the absence of protective neurosensory tissue, has direct implications for surgical planning. Awareness of this anatomical risk may prompt surgeons to adopt protective intraoperative strategies such as minimizing illumination intensity and duration, limiting repeated dye staining, or using perfluorocarbon liquid or viscoelastic barriers during BBG application to help preserve foveal integrity and optimize visual outcomes.

CONCLUSION

BBG-related retinal phototoxicity may occur following ILM peeling, with eyes undergoing MH surgery appearing more vulnerable due to the absence of protective foveal neurosensory tissue. These findings suggest that underlying retinal architecture may influence susceptibility to dye- and light-related injury. Given the retrospective nature of this case series, the observations should be considered hypothesis-generating, and prospective studies incorporating functional and microvascular assessments are required to confirm these associations. Awareness of this potential risk may help guide surgeons toward adopting protective intraoperative strategies to preserve foveal integrity and optimize visual outcomes.