Displacement of Submacular Hemorrhage Using Subretinal Cocktail Injection versus Pneumatic Displacement: A Real-World Comparative Study

Submacular hemorrhage (SMH) is a vision threatening complication of neovascular age-related macular degeneration (nAMD) and polypoidal choroidal vasculopathy (PCV). Without prompt treatment, it could cause irreversible photoreceptor damage and vision loss within 2 weeks through the combined effects of physical barrier impairing metabolic exchange, mechanical shearing forces, and iron toxicity [1‒3]. The drop in visual acuity (VA) at diagnosis of SMH is profound, and the visual recovery is limited despite intravitreal anti-vascular endothelial growth factor (VEGF) agent injection. At 1 year, more than half of the cases lost at least 3 lines of vision and less than a quarter regained their vision prior to SMH diagnosis [4].

While SMH only affects 1.37% of nAMD patients [4], it is much more common in PCV as 11.1% of PCV patients developed SMH within 5 years of diagnosis [5, 6]. The cumulative incidence of massive SMH (larger than 4 disc diameter) is approximately three times higher in PCV than nAMD [7]. The proportion of PCV the Asian patients with nAMD phenotypes is estimated to be between 20 and 60% based on indocyanine green angiography (ICGA), which is much higher than the Caucasian population (8–13%) [8, 9]. Hence, the incidence of massive SMH is likely higher in the Asian population due to the higher prevalence of PCV.

Early surgical intervention to displace SMH from the fovea could potentially preserve central vision and avoid permanent photoreceptor damage, especially in cases with thick and large SMH where the effect of anti-VEGF monotherapy may be limited [5]. Pneumatic displacement with intravitreal expansile gas and pars plana vitrectomy (PPV) with subretinal injection of tissue plasminogen activator (tPA) are the mainstays of surgical options. The success rate of pneumatic displacement with or without tPA is variable [10‒12]. In addition, tPA molecules in the vitreous cavity may not reach the target subretinal blood clot due to its molecular size [13]. In fact, intravitreal tPA was shown to have additional benefit to pneumatic displacement alone in one study [12]. Conversely, subretinal injection of tPA was reported to have a higher displacement rate than intravitreal injection [14]. Recently, Martel and Mahmoud proposed subretinal injection of a cocktail solution consisting of tPA, bevacizumab, and filtered air following PPV to improve displacement rate [15], with a complete foveal displacement rate ranges from 92 to 100% [16‒19]. There are, however, scarce data on the safety and efficacy of this technique in small-sized SMH [20].

To date, there is no consensus on the optimal surgical technique to displace SMH as there is a paucity of studies comparing different surgical techniques [20]. Moreover, most of these studies did not include eyes with PCV and extensive SMH that extends beyond temporal vascular arcade [21, 22]. There is lack of head-to-head comparison between pneumatic displacement and the novel subretinal cocktail injection, particularly in the Asian population where PCV prevalence is high and SMH is more common. In this retrospective interventional case series study, we compared the surgical outcomes and safety profiles of pneumatic displacement versus PPV and subretinal cocktail injection in treating Asian patients with SMH.

This is a multicenter retrospective interventional case series performed at the Hong Kong Eye Hospital (HKEH) and Prince of Wales Hospital (PWH), Hong Kong. The study was approved by the Local Ethical Committee and adhered to the tenets of the Declaration of Helsinki. An electronic medical system was used to identify patients who underwent surgical SMH displacement secondary to nAMD or PCV from May 1, 2015, to October 31, 2022. Patients underwent either pneumatic displacement with intravitreal expansile gas or PPV with subretinal cocktail injection. The choice of operation to displace SMH was determined by operating surgeons’ discretion. In eyes with cataract who underwent PPV with subretinal injection, concomitant phacoemulsification would be performed. The presence of cataract alone would not influence the choice of surgical technique. For patients who underwent pneumatic displacement as primary operation, if the SMH was not successfully displaced by 1 week, a secondary operation with PPV and subretinal injection would be considered based on clinicians’ discretion. Patients with the following criteria were excluded: (1) Lost to follow-up or lack of clinical data up to 12 months; and/or (2) SMH due to causes other than nAMD or PCV, such as myopic choroidal neovascularization and ruptured retinal artery macroaneurysm.

In the pneumatic displacement group, patients underwent intravitreal injection of expansile gas of either 0.3 mL pure perfluoropropane gas (C3F8) or 0.5 mL pure sulfur hexafluoride gas (SF6) with or without concomitant intravitreal injection of tPA and anti-VEGF agent (ranibizumab or aflibercept). The choice of C3F8 or SF6 was made by operating surgeon’s discretion. Anterior chamber paracentesis was performed as required to normalize the intraocular pressure (IOP). In general, C3F8 was preferred by most surgeons because of the smaller initial volume of injection required. This preference is based on our experience that multiple attempts of anterior chamber paracentesis are required to normalize IOP in Asian eyes when a larger volume of intraocular gas was injected. Patients were instructed to maintain a prone or face-down posture for 1 week postoperatively.

In the PPV with subretinal cocktail injections group, 23-gauge PPV was performed (Constellation, Alcon, Fort Worth, TX, USA), and posterior vitreous detachment induction or internal limiting membrane peeling would be performed in selected case to assist subretinal injection. This is followed by subretinal injection of a cocktail solution containing either 2 mg/0.05 mL aflibercept (Eylea, Bayer, Leverkusen, Germany) or 0.5 mg/0.05 mL of ranibizumab (Lucentis, Novartis, Basel, Switzerland) and 50 μg/0.4 mL of tPA (Activase, Genentech, CA, USA). Subretinal injection was performed with an extendible 41-gauge (G) subretinal needle which has a 23-gauge/0.6-mm outer diameter (DORC International, The Netherlands). This was followed by subretinal injection of 0.2 mL of filtered air. A partial fluid air exchange of approximately 50% of vitreous cavity was then performed, which is subsequently exchanged with 20% sulfur hexafluoride (SF6) (online suppl. Video 1; for all online suppl. material, see https://doi.org/10.1159/000537953). The surgery would be combined with phacoemulsification and intraocular lens implantation in eyes with visually significant cataract. Patient was instructed to maintain an upright posture for at least 5 days after the operation.

Injection site was chosen to avoid pigmented epithelial detachment (PED) and away from fovea, near inferotemporal arcade over area of most bullous subretinal fluid or hemorrhage, where the subretinal space is expanded and distance between retina and retinal pigmented epithelium (RPE) is increased. As soon as retinal surface dimple is made by the 41-G cannula, cocktail solution or air was slowly injected under low pressure, using the jet of fluid or air to enter the subretinal space. The advancement of 41-G cannula would be stopped immediately at the first sight of subretinal bleb. Intraoperative OCT was not used as it was not available.

In both the pneumatic displacement and PPV with subretinal cocktail injection groups, postoperative intravitreal anti-VEGF injections was performed on a pro re nata (PRN) basis, and verteporfin photodynamic therapy was offered to patients with a diagnosis of PCV.

Patients’ demographics, medical, and ophthalmic history were obtained from medical records. The duration of symptom before presentation and prior history of anti-VEGF treatment were documented. VA was recorded with Snellen chart and converted to logarithm of minimum angle of resolution (logMAR) scale for statistical analysis, visual acuities of finger count and hand movements were converted to 2.0 and 3.0 logMAR or 0.014 and 0.005 Early Treatment for Diabetic Retinopathy (ETDRS) letters score, respectively [23, 24]. Complete ophthalmic examination was performed, including IOP measurement, anterior segment and dilated fundus examination were performed by retinal specialists at baseline, 1 week, and 1, 3, 6, and 12 months postoperatively. Spectral domain optical coherence tomography was performed at baseline and postoperatively with either SPECTRALIS (Heidelberg Engineering, Heidelberg, Germany) or Cirrus (Carl Zeiss Meditect, Dublin, CA, USA). Fundus fluorescein angiography and ICGA were performed (Heidelberg Engineering, Heidelberg, Germany) postoperatively to determine the underlying cause of SMH (Fig. 1).

The SMH extent was graded according to the system proposed by Sharma et al. [18]. A SMH is considered small if it lies within the vascular arcade, large if it reaches the arcades, extensive if it extends beyond the arcades, and massive if it extends beyond the equator, involving 2 retinal quadrants or both (Fig. 2a, 3a). The level of SMH was classified as purely subretinal hemorrhage, combined subretinal and sub-RPE hemorrhage with predominantly subretinal component, combined subretinal and sub-RPE hemorrhage with predominantly sub-RPE component and purely sub-RPE component based on OCT (Fig. 2b, 3b). The central subfield thickness (CST) was automatically generated by in-built software of OCT devices using the volume scan.

Degree of SMH displacement at 1 week postoperatively was graded as (1) displacement to beyond equator, (2) beyond vascular arcades, (3) within arcades but out of fovea, and (4) persistent subfoveal hemorrhage (Fig. 2c, 3c). Postoperative complications including macular hole (MH) formation, retinal detachment (RD), vitreous hemorrhage (VH), hyphema, IOP spike, and recurrence of SMH were documented.

All ophthalmic images, including grading of SMH extent and level, were performed by S.K.H. and M.B., both are retinal specialists with at least 3 years of experience after obtaining fellowship in vitreo-retina training.

Baseline characteristics and postoperative outcomes were compared between the two surgical groups. Continuous variables, such as VA and CST, were compared using analysis of covariance adjusted for covariates. Variables are included as covariates in the analysis of covariance analysis if they have p value <0.1 on univariate regression analysis. Based on this criterion, age, baseline VA, AMD subtype, prior anti-VEGF treatment, extent of SMH, duration of symptoms before surgery, and presence of cataract were selected as covariates. Mena change in postoperative VA was reported in ETDRS letters equivalent, Wilcoxon signed rank test was used to compare postoperative with baseline ETDRS VA.

Categorical variables, such as extent of SMH displacement, were tested with Pearson χ² test. Fisher’s exact test would be used instead when the assumptions of χ² test were not met. To avoid outliers affecting the comparison, eyes with postoperative RD and preexisting fovea involving macular scar were excluded in the comparison of postoperative VA.

Multivariate regression analysis was performed to identify predictors of postoperative VA and successful SMH displacement. Linear regression was used for continuous variable outcome. Logistic regression was used when the outcome was categorical variable, such as rate of SMH displacement, and the odds ratio would be reported. Subgroup analysis was performed to compare the baseline and postoperative clinical characteristics of the two AMD subtypes. All statistical analysis was performed using SPSS software version 25 (IBM).

This study included 63 eyes from 63 patients, of which 40 (63.5%) and 23 (36.5%) eyes underwent pneumatic displacement and PPV with subretinal cocktail injection as primary procedure, respectively. Patients’ demographics, diagnosis, and clinical characteristics are summarized in Table 1.

There was no statistically significant difference in patients’ demographics between the two groups, such as mean age and gender. However, the pneumatic displacement group had larger proportion of eyes with PCV (72.5% vs. 43.5%, p = 0.032), less extensive SMH (p = 0.005), shorter duration between symptom onset and surgery (5.46 vs. 10.65 days, p =<0.001), and thinner baseline CST (780.2 μm vs. 1,006.6 μm, p = 0.012). There was no significant difference in baseline VA, cataract status, and number of prior anti-VEGF injections between the two surgical groups.

The surgical, postoperative functional, and anatomical outcomes are summarized in Table 2. In the pneumatic displacement group, 38 (85%) received intravitreal C3F8 and 2 (5%) received SF6. 8 (42.1%) and 5 (12.5%) eyes underwent concomitant intravitreal anti-VEGF and tPA injection, respectively. In the PPV and subretinal injection group, 14 (73.7%) and 4 (26.3%) eyes received subretinal aflibercept and ranibizumab, respectively. Three (33.3%) patients in the pneumatic displacement group did not receive postoperative anti-VEGF injection in view of poor visual prognosis from extensive macular scar. The mean number of anti-VEGF within the first year was similar between the two groups (p = 0.305), which was 3.10 and 2.70 for the pneumatic displacement and subretinal injection group, respectively. Postoperative verteporfin photodynamic therapy was more commonly performed in the pneumatic displacement group, but the difference was not statistically significant (25.0% vs. 8.7%, p = 0.183).

The subretinal cocktail injection group was more 11.1-fold more likely to achieve complete foveal displacement (87% vs. 40%, p < 0.001, OR = 11.1) and 5.15-fold more likely to achieve displacement to arcade or beyond (52.5% vs. 17.5%, p = 0.009, OR = 5.15) compared to pneumatic displacement. In addition, greater reduction in CST was achieved in the subretinal injection group.

After adjusting for covariates mentioned before, the early postoperative VA was similar in both groups at 1-, 3-, and 6-month visits. At 12-month follow-up, the pneumatic displacement group had significantly better VA (0.64 vs. 1.03, p = 0.040).

In the pneumatic displacement group, the mean VA improved from 1.46 at baseline to 1.17 at 1 month, 0.90 at 3 months, 0.67 at 6 months, and 0.64 at 12 months visit. In the subretinal injection group, the mean VA improved from 1.62 to 1.25 at 1 month, 0.66 at 3 months, 0.91 at 6 months, and 1.03 at 12 months visit.

The serial changes in postoperative ETDRS letter score VA are presented in Table 3, and box plot of postoperative VA over time was presented in Figure 4. Significant improvement in VA compared to baseline was observed in both surgical groups at 1-, 3-, and 6-month follow-up visits. At the 12-month follow-up, only the pneumatic displacement group demonstrated significant improvement in VA compared to baseline but not the PPV with subretinal cocktail injection group.

At least 10 letters gain in VA were achieved in both groups up to 12 months. The mean change in VA compared to baseline at 1 month was +15.0 (p = 0.001) and +12.0 (p = 0.011) letters, at 3 month +22.1 (p < 0.001) and 22.2 letters (p < 0.003), at 6 months +26.8 (p < 0.001) and +14.8 (p < 0.038) letters, at 12 months +27.7 (p < 0.001) and +11.5 (p = 0.062) letters for pneumatic displacement and PPV with subretinal cocktail injection group, respectively.

The rates of postoperative complications, including VH, hyphema, IOP spike, MH, retinal break, RD, and need for reoperation, are summarized in Table 4. There was no statistically significant difference in the rate of postoperative SMH recurrence, VH, hyphema, IOP spike, or RD.

The pneumatic displacement group had higher reoperation rate (20% vs. 12.0%), but the difference was not statistically significant (p = 0.732). In the pneumatic displacement group, 4 cases required PPV for non-clearing postoperative VH, and 1 case underwent pneumatic retinopexy with cryotherapy for RD. One patient developed postoperative breakthrough VH and total RRD that underwent PPV but re-detached afterward. Two patients with failed pneumatic displacement received subretinal cocktail injection as a second operation, after which foveal displacement was achieved.

In the PPV with subretinal cocktail injection group, 1 patient developed postoperative MH which was successfully closed with reoperation with ILM peeling and gas tamponade. One patient had postoperative persistent VH with hyphema and IOP elevated to 30 mm Hg that could not be controlled medically and needed PPV with anterior chamber washout. The IOP normalized after the reoperation, and no further glaucoma operation was required. One patient had recurrence of SMH with pneumatic displacement performed 3 months after the primary operation.

Although the rate of postoperative RD was similar between the two groups, the RD in the subretinal injection group tended to be more severe. Two patients developed total RD in the subretinal injection group with final VA of light perception. The two eyes had either extensive or massive SMH on presentation. RD repair surgery was not offered in these patients in view of poor prognosis. In the pneumatic group, 1 patient had limited superior RD which was successfully repaired with pneumatic retinopexy with cryotherapy, the final logMAR VA was 0.52. Another patient had postoperative breakthrough VH and total RD, the retina re-detached despite two reoperations with PPV, this patient’s final VA was hand movement.

The results of regression analysis for predictors of baseline and postoperative VA were summarized in online supplementary Table 1. Older age was associated with worse baseline VA and postoperative VA at 6 and 12 months. Extensive SMH was associated with worse baseline VA only but not with postoperative VA. Interval between symptom onset and operation was associated with worse postoperative VA at 6 and 12 months. AMD subtype and concomitant cataract surgery were not associated with postoperative VA.

The comparison of the two AMD subtypes was summarized in online supplementary Table 2. More patients with nAMD underwent subretinal cocktail injection compared to PCV (54.2% vs. 25.6%, p = 0.022). There were more eyes with SMH reaching arcade or beyond in eyes with PCV (46.2%) compared to nAMD (37.5%), although this difference was not statistically significant (p = 0.480). The baseline and postoperative VA and CST were similar in both groups. Complete foveal displacement at 1 week was significantly more common (p = 0.015) in the nAMD group (37.5%) compared to the PCV group.

In this study, we described the efficacy and safety of SMH displacement by PPV with subretinal cocktail injection and compared it with pneumatic displacement. Despite the subretinal injection group had significantly more extensive and thicker SMH on presentation and longer duration of symptom before operation, similar postoperative VA was achieved compared to pneumatic displacement up to 6 months. The likelihood of achieving complete foveal displacement and SMH displacement to arcade or beyond using subretinal cocktail injection was 11.1 and 5.15 folds higher than pneumatic displacement, respectively. The CST reduction was also greater in the subretinal injection group, although this may be explained by the higher baseline CST in the subretinal injection group. Moreover, clinically meaningful improvement in vision (>10 ETDRS letters improvement) compared to baseline was achieved in both surgical groups up to 12 months, providing new evidence to support early surgical displacement of SMH. Retinal surgeons may take these results into consideration in managing patients with SMH.

The pneumatic displacement technique only achieved complete foveal displacement in 37.5% of eyes in our case series; two of these failed cases underwent successful SMH displacement with subretinal cocktail injection as a secondary “rescue” operation. In terms of postoperative VA, there was no significant difference between the two surgical techniques up to 6 months in our cohort despite the subretinal cocktail injection group had significantly more extensive SMH at diagnosis, longer duration of symptom before surgery, and thicker baseline CST. Suggesting the subretinal injection technique may have functional benefit over pneumatic displacement, at least in the early postoperative period, in extensive SMH and delayed presentation. Therefore, subretinal cocktail injection can be considered in patients with failed pneumatic displacement, extensive SMH, or delayed presentation. In fact, a group from the UK recently proposed a stepwise approach in SMH displacement. According to this sequential approach, eyes with SMH were first treated with intravitreal expansile gas and tPA, patients with failed SMH displacement on postoperative day 3 would be undergo PPV with subretinal injection. With this protocol, they achieved 87.1% and 60% successful displacement rates in the whole group and the reoperated patients, respectively [25].

Clinically meaningful improvement in VA (at least 10 letters gain in ETDRS letter score) was achieved and maintained in both surgical groups up to 12 months postoperatively. However, in the subretinal injection group, this improvement in VA declined and became statistically insignificant at 12-month follow-up (p = 0.062) despite there was a mean gain of 11.5 letters. It is also worth noting that at 12-month visit, the subretinal injection group had slightly worse vision compared to the pneumatic displacement group. This could be explained by the higher rebleeding rate in the subretinal injection group and overall relative undertreatment with postoperative anti-VEGF injections in our study. The late decline of VA highlights the importance of continued anti-VEGF injections to maintain long-term VA and prevention of rebleeding following successful surgical SMH displacement [20, 26, 27].

The subretinal cocktail injection may increase SMH displacement rate through several mechanisms. First, subretinal tPA directly reaches the target site, while intravitreal tPA molecules may not be able to penetrate to the subretinal space [13]. This advantage of subretinal tPA may be even more important in patients who presented late, as the subretinal cocktail injection group achieved similar VA despite it had significantly longer duration of symptoms before surgery. Although there are concerns regarding the safety of subretinal tPA based on animal studies [28, 29], most reports in the literature showed improvement in vision and retinal sensitivity measured by microperimetry when a dosage of 12.5–50 μg was used [14, 18, 27, 30, 31]. The “cocktail” combination of subretinal tPA with subretinal anti-VEGF promotes inferior displacement of SMH by increasing subretinal volume and reducing mechanical friction [15]. The subretinal air bubble reduces the buoyancy of subretinal blood by as much as 830 folds [32], enhances oxygenation to photoreceptors, and acts as an air lock to allow mixing of the subretinal tPA and anti-VEGF agents [18]. The subsequent gas in the vitreous cavity promotes inferior displacement of SMH and prevents superior migration of subretinal air [15]. Moreover, patients can adopt a postoperative upright posture, avoiding the need of prolonged face down or prone posture after the operation.

Apart from promoting SMH displacement, subretinal anti-VEGF injections may have additional benefits. First, it is possible that the vitreous half-life of anti-VEGF agents is reduced in vitrectomised, gas-filled eye [33, 34], and subretinal penetrability may be decreased in massive SMH [26, 35]. Moreover, subretinal aflibercept was associated with a lower need for subsequent anti-VEGF injection compared to intravitreal aflibercept injection over a 24-month of follow-up period in a recent study, this may be explained by higher subretinal concentration of anti-VEGF agent [36]. There are concerns, however, about the compatibility of aflibercept with tPA. A study showed that aflibercept is cleaved by plasmin, a product of tPA, while ranibizumab and bevacizumab are not. However, aflibercept’s VEGF suppression is not affected at a clinical concentration [37]. The co-application of tPA and aflibercept may be justified, particularly in the Asian population where PCV is a more common cause of massive SMH [5, 8] and aflibercept has a higher rate of polypoidal lesion regression compared to ranibizumab [38].

Despite the potential benefits of subretinal cocktail injection, there are caveats associated with this procedure. First, subretinal injection is surgically challenging and requires the use of specialized instrument such as the 41-G microneedle. Careful surgical planning is required to prevent inadvertent puncture of RPE, which may lead to RPE rip [26, 35]. Ideally, subretinal injection should be performed near inferior arcade, preferably area with bullous hemorrhagic displacement of the retina, to promote inferior displacement of subretinal clot while avoiding area with PED. Hence, intraoperative OCT, if available, may assist retinal surgeons in making more precise and safer subretinal injection. Second, iatrogenic MH is a unique complication associated with subretinal injection but not with pneumatic displacement, as reported in our case series and previous publications [18, 30]. Iatrogenic MH may occur as a result of increased submacular hydrostatic pressure resulting from subretinal injection which may be minimized by performing subretinal injection under a perfluorocarbon liquid bubble [39]. Lastly, from our experience, when postoperative RD occurred after subretinal cocktail injection, it is frequently inoperable with a final VA of light perception. Both eyes had either extensive or massive SMH on presentation, the risk of vision threatening RD may be higher in eyes with extensive SMH. Patients should be informed of the risk of these complications before undergoing surgery.

Massive SMH is more common in PCV compared to nAMD [5], but PCV may have better visual prognosis overall due to diminished CNV activity after rupture of polypoidal lesions and less subretinal fibrosis [40]. Conversely, sub-RPE hemorrhage is more common in PCV [41], which was believed to prevent successful SMH displacement in the past [42], but recent reports showed that sub-RPE hemorrhage could also be effectively displaced by subretinal tPA with or without anti-VEGF and air [18, 43, 44]. In our cohort, although extensive SMH was more common in PCV eyes, the difference was not statistically significant. In addition, there were no significant differences in postoperative VA between the AMD subtypes. Complete foveal displacement was more common in the nAMD group, but this could be confounded by the fact that nAMD underwent subretinal injection more commonly than PCV patients in our study. It is important to note that the prevalence of PCV could be underestimated in our cohort, particularly in the subretinal injection group, as polyp regression could occur after anti-VEGF therapy or rupture of polypoidal lesions [40], and ICGA was performed after resolution of SMH in the current study.

To the best of our knowledge, this is one of the first studies to compare the surgical outcomes of pneumatic displacement versus subretinal injection of cocktail solution consisting of tPA, anti-VEGF agent, and air. We demonstrated that subretinal cocktail injection has a good displacement rate from small to massive SMH. Our study provided new data to inform retinal surgeons on the decision of surgical technique for management of SMH. As subretinal cocktail injection achieved higher rate of successful displacement of SMH despite more extensive baseline bleeding and longer symptom onset, it may be reasonable to consider this surgical technique in patients with extensive SMH or delayed presentation if surgical expertise and equipment are available. Moreover, it can be used as a “rescue” secondary operation in cases with failed displacement with intravitreal expansile gas alone.

This study has several limitations. First, it is a retrospective study with a relatively small sample size, and the two surgical groups differed significantly in some baseline characteristics. Notably, patients in the subretinal injection group presented later and had more extensive SMH and greater CST at baseline and less proportion of PCV, and these baseline differences may explain the higher CST reduction in the subretinal injection group. We tried to adjust for the difference in baseline characteristics by adjusting confounding factors in the statistical analysis as much as possible. Nonetheless, the subretinal injection group achieved similar postoperative VA up to 6 months and higher displacement rate despite more severe SMH at baseline and longer duration of symptoms before operation. Secondly, the subretinal cocktail injection group had significant longer interval between symptom onset and operation. This can be explained by the fact that subretinal injection requires expertise from experienced surgeons and specialized surgical equipment, which may not be readily available after office hours. Conversely, pneumatic displacement can be performed safely by less experienced surgeons and requires basic surgical equipment only. Thirdly, there could be selection bias as retinal surgeons may prefer subretinal cocktail injection for more extensive SMH and late presenters. Fourthly, only 12.5% patients in the pneumatic group received intravitreal tPA, and 10% did not receive concomitant or postoperative anti-VEGF injections, making direct comparison with subretinal cocktail injection difficult. Lastly, there was no standardized postoperative anti-VEGF injection protocol and relative undertreatment using anti-VEGF therapy, which may affect postoperative vision and rebleeding rate [26]. It is also worth noting that there is no consensus on the definition of successful SMH displacement; hence, direct comparison of SMH displacement rates among studies should be avoided. These limitations can be rectified by future prospective randomized controlled trials with standardized postoperative anti-VEGF injection protocol and standardized reporting of SMH extent and displacement. Microperimetry and multifocal electroretinogram may also be included in future studies to quantify postoperative retinal sensitivity.

Our study showed that both pneumatic displacement and PPV with subretinal cocktail injection could lead to clinically meaningful improvement in VA. But the subretinal cocktail injection had a higher successful SMH displacement rate despite patients had more extensive SMH and presented later in this group. Moreover, it could successfully displace SMH in cases when pneumatic displacement as primary surgery failed. We provided new data to inform retinal surgeons in the decision-making of surgical management of SMH. Further studies on PPV with subretinal injection of tPA, anti-VEGF, and air are warranted.

The authors have no relevant acknowledgment.

This study was conducted in accordance with the principles of the Declaration of Helsinki. It was approved by the Research Ethics Committee (REC) of the Kowloon Central Cluster, Hospital Authority, Hong Kong SAR. Reference no.: KCC/KEC-2020-0028. Written informed consent was obtained from all patients prior to the surgical procedure.

The authors declare no conflicts of interest.

There is no funding source to declare.

Simon K.H. Szeto designed the study, performed statistical analysis, and prepared the first draft of the manuscript. Chi Wai Tsang, Shaheeda Mohamed, Timothy Y.Y. Lai, Gary K.Y. Lee, Jerry K.H. Lok, Li Jia Chen, and Marten Brelen revised the manuscript. Vivian W.K. Hui and K.K. Tsang collected the data. All authors contributed to the study concept and design and read and approved the final manuscript.

All data generated or analyzed during this study are included in this article. For any inquiry, please contact the corresponding author.