The current meta-analysis of five RCTs showed that the overall MH closure rate in the FDP group was significantly higher than that in the nFDP group. The subgroup meta-analysis also demonstrated significant differences in the MH closure rate between the two groups with MH sizes larger than 400 μm but not with MH sizes smaller than 400 μm. Our meta-analysis strongly suggests that the success rate of MH healing is related to the face-down position when MH size is larger than 400 μm. Therefore, it is necessary to keep these patients face down after MH surgery with gas tamponade. However, for patients with MH sizes smaller than 400 μm, postoperative face-down posture is not mandatory, as long as supine positioning is avoided. This finding will help individuals with small MHs (≤ 400 μm) who are hesitant of the surgery because of the postoperative prone position constraint seek surgical repair for MH. Furthermore, the adoption of a non-face-down position may spare patients painful discomfort and avoid certain complications, and in certain cases, shorten the duration of hospitalization or absence from work . For those scholars who think postoperative non-supine position adoption is adequate for all MH patients after surgery [9, 10, 23], we suggest that the position required according to the MH size is more appropriate rather than blindly abandoning the postoperative position.
The formation of MHs is caused by the anteroposterior and tangential traction from posterior vitreous detachment and the proliferation of glial cells [24, 25]. The traction can be removed by pars plana vitrectomy with ILM peeling . In the case of gas filling, a MH smaller than 400 μm was enough to achieve successful anatomic healing without the need for the prone position . This finding may depend on the hydration theory that the subfoveal subretinal fluid is pumped out by the retinal pigment epithelium quickly, once traction is relieved . However, for MHs larger than 400 μm, although the tangential force was removed by ILM peeling, the MH was also difficult to heal. Because the distance between the broken ends of the retina is so large, it requires a strict prone position to ensure that the gas bubble, under a sufficient or partial absorption state, keeps the intraocular gas-macula contact extended, drawing the edges of the hole into apposition with each other, thus providing a scaffold for the migration of glial cells, blocking fluid entry into the hole and moving the subretinal fluid to reattach the retina. These processes contribute to successful anatomic healing [8, 21, 29]. This may explain why, for patients with large MHs, the successful healing rate of the FDP group was higher than that of the nFDP group. No significant difference was found in the MH closure rate of patients with small MHs between the FDP and nFDP group.
Furthermore, the gas will be gradually absorbed with time, and the buoyancy will be weakened after partial gas absorption. Within 3 days, 30% of the gas will be absorbed , which does not affect the healing of MHs smaller than 400 μm. This may be due to the removal of the traction force by ILM peeling which is important for hole healing in addition to the top pressure effect of the gas. However, for the larger holes, the traction force removal is not enough to make the hole close gradually. Gas-macula contact and upward mechanical force on the macula are required for effective tamponade. The retina must be acted on by an external force for it to be displaced, thus apposing the MH edges . Partial gas absorption with time will affect the buoyancy, and therefore the face-down position is needed to keep the remaining gas buoyant against the MH and to sustain apposition of the gas bubble to the inner surface of the MH to facilitate closure. In other words, the upward mechanical force of the gas on the macula, blocking the action of fluid entry into the hole and providing a scaffold/bridge for gliosis play an important role in the healing process of large holes.
Previous studies have reported that the healing of MHs begins within 24 h after the surgery, and the bridge configuration occurs around 3 days thereafter [30, 31]. The MHs were basically healed within 3 days after surgery, and those that were not healed within 3 days were still open during the 3-month follow-up [3, 21]. All of the patients from the studies included in this meta-analysis were in the prone position for 3 days or more, so the comparison of the closure rate of the MHs with sizes larger than 400 μm in the FDP and nFDP groups after MH surgery is credible. In short, patients with MHs larger than 400 μm should keep the face-down positioning strictly for 3 days after MH surgery. For patients who MH closure is not achieved within 3 days in an FDP, continuing an FDP will not increase healing. A strict prone position for 3 days after the operation is not a long time and should have little impact on systemic complications, which is conducive to improving the compliance of patients.
The studies included in this meta-analysis did not provide the same FDP time following surgery. It should be noted that because many MH patients are elderly, compliance to the prone position is usually not high due to systemic factors, and therefore, the five included studies only required patients to keep the FDP for a few hours per day. In addition, although patients are advised to remain face down everyday by their surgeon, they cannot be supervised after they leave the clinical setting. Hence, there may be some patients showing low compliance who did not spend sufficient time in the FDP to meet treatment requirements, which may have affected the accuracy of the study results. As there is currently no consensus on the number of hours necessary for patients with larger MHs to remain in the FDP daily, RCT studies with large sample sizes investigating this topic are necessary in future. In addition, although MH duration prior to surgical intervention is an important variable affecting closure rate, there was no significant difference in MH duration between the FDP and nFDP groups in the included studies; hence, we believe that MH duration did not influence the results of our meta-analysis.
There were some limitations to this study. First, all of the included studies had small sample sizes, and there is a lack of data with large sample sizes from multi-center RCTs in our research. Second, this meta-analysis was restricted to studies published in indexed journals and we did not search for unpublished studies, original data, or papers published in non-English languages, which may have led to publication bias. Meanwhile, two studies [18, 19] reported 100% MH closure rate in the FDP and nFDP groups for MH sizes smaller than 400 μm, which resulted in a reduction in the number of publications included in the bias analysis. This methodology resulted in an overemphasis on positive results and neglect of negative results, potentially accounting for the publication bias described. Third, although Zhang et al. used the inverted ILM flap technique when performing surgery on MHs larger than 400 μm, the rate of use of the inverted ILM flap technique was not significantly different between the FDP and nFDP groups. However, 100% MH closure was achieved in patients undergoing the inverted ILM flap technique with and without face-down posturing. Therefore, although we believe that the use of different surgical techniques did not affect the results of our meta-analysis, it remains a possible source of bias. Fourth, we did not perform subgroup meta-analysis for visual acuity between the two groups due to the different definitions of visual acuity improvement among the five included studies. The images of the repaired MHs were categorized into three patterns (U-type, V-type, and W-type) by optical coherence tomography . We were also not able to perform a subgroup analysis of the closure type due to lack of relevant data. Lastly, the area of ILM peeling, air or longer-acting gas tamponade and whether the lens were removed may also have influenced the closure rates . However, we were unable to find sufficient data to investigate these parameters. Therefore, longitudinal in-depth RCTs with large sample sizes evaluating the aforementioned parameters are required in future to investigate differences between the two groups.