Our study demonstrated that the single-operation success rate of PPV with air tamponade for the 1715 patients with RRDs was 92.0%, which is in tandem with previous studies where the primary reattachment rate varies from 79.0% to 100.0% [1, 4,5,6,7,8,9,10,11,12,13,14]. We analyzed the failure causes and described their distribution. Consistently, the leading cause was new or missed breaks, which accounted for nearly half of the failure cases (48.9%) [1, 6, 7, 10,11,12,13,14,15,16]. Breaks reopening comprised the second frequent cause, accounting for 43.8%, followed by PVR (7.3%).

Undetected breaks or formation of new breaks, and progressive PVR after vitrectomy have been well characterized [1, 17]. However, in most reported cases of reopened breaks, no definite cause was allocated. In this study, patients with large retinal break(s) (≥ 1.5 DD) had higher chances of breaks reopening. From our intraoperative observation during salvage vitrectomy, incomplete adhesion of large breaks, and most of the chorioretinal scar failed to develop at the anterior border. This finding is parallel with Tornambe et al. who observed that large tears up to 2.5 clock hours in size was a predictor for failure in pneumatic retinopexy [21].

Importantly, there was a trend where the median duration of symptoms for eyes with reopened break (11.5 days) was longer than those without (7.0 days), but it did not reach statistical significance. This finding is compiled with our clinical notions that long-lasting retinal detachment is a presumed risk factor for surgery failure for a higher grade of preoperative PVR.

From our observation, the median time to recurrence caused by reopening of retinal breaks (18.0 days, IQR: 12.0–25.0 days) was remarkably shorter than those without breaks reopening (28.0 days, IQR: 13.0–58.5 days). Although numerous studies have suggested that adhesion between the neurosensory retina and the retinal pigment epithelium (RPE) developed within 24 h of vitrectomy [22,23,24]. Postoperatively, with proper prone positioning, air tamponade closed retinal breaks by its surface tension and buoyancy [25]. However, the adhesion formed in the early days may not be sufficiently strong to seal the retinal edge around the tear, indicating that time is required for permanent scar formation [23, 24]. The shear stress at the circumference of breaks leads to its reopening. This stress is maximum with sudden jerking head movements and greatest at the fluid-air interface [26]. Also, with the head movements, residual SRF could shift, especially to the inferior breaks, in such cases, shifted SRF could cause reopening of treated retinal breaks in the early postoperative period. In this study, all the patients were instructed to maintain a proper head positioning for 1 week and have restricted activities for 1 month. However, in the follow-up interview, most of them completed prone positioning but could not adhere to restricted activities for 1 month. The cases with reopened breaks had significantly less time of restricted activities than the those without. As the air tamponade disappeared within the 2 weeks after vitrectomy [27], most of patients return to work at this time for rapid visual rehabilitation with air tamponade. However, among 60 patients with retinal breaks reopening, 33% of them relapsed within 2 weeks postoperatively, and half of them recurred at 3–4 weeks. Therefore, instead of emphasizing the prone positioning, restricted activities should also be considered a critical factor. In agreement with previous findings, Martínez-Castillo et al. yielded favorable success rate for RRD with inferior breaks repaired by PPV with gas tamponade after 24-h facedown position compared with no prone position [13,14,15]. This series study demonstrated that facedown position did not influence the development of a chorioretinal adhesion in the treated retinal tears. Further, we would not recommend the avoidance of long-acting gas tamponade in all patients, especially for the patients with poor compliance since an intraocular gas bubble may act as a reminder that excess and vigorous head movements should be avoided. An animal study revealed that it was not until 4 months after surgery that the regenerated RPE cells presented with the same morphological characteristics as normal RPE cells. Fluorescein angiography showed no leakage at the original RPE wound by 1 month postoperatively, indicating that the blood-retinal barrier was reconstructed by the regenerated RPE cells [28]. Although most authors did not indicate whether patients maintained restricted activities or not, or how long it lasted, we recommend prolonged period for restricted activities to bring the retina and choroid in contact for sufficient time to produce a firm chorioretinal adhesion to permanently close the break and allow the SRF to be completely cleared by the RPE.

In addition, traction by the remnant vitreous near the edge of retinal tears may reopen sealed breaks. In the salvage vitrectomy, a thin and condensed vitreous was observed at the retinal breaks in most cases. Conformed to previous publications, vitreoretinal traction remains an important cause of breaks reopening [29]. Vitreoretinal surgical expertise is demanded to remove vitreous completely around retinal breaks and in the periphery at the vitreous base. This can be achieved by doing total PPV and 360° vitreous base shaving under prism lens with higher magnification. Triamcinolone was used to visualize the vitreous (Additional file 1). From our observation, a thorough vitreous cortex removal is beneficial for preventing recurrence. Parallel findings were seen in previous studies where the vitreous cortex remnants (VCR) increased the recurrence of retinal detachment due to persistent tangential traction to form new breaks or reopen preexisting breaks [30]. In addition, some authors have suggested that peripheral VCR induced PVR [31], and VCR at the macula may act as a scaffold for fibrocellular proliferation [32]. Therefore, adequate triamcinolone-assisted VCR removal at both posterior pole and midperipheral should be performed to prevent recurrence.

This report analyzed the causes of breaks reopening in an attempt to highlight the important preoperative, operative, and postoperative factors that can contribute to redetachment, and by taking prophylactic measures, the surgical outcome can be improved. During preoperative assessment, it is crucial to examine the detached retina to locate the breaks and zones with vitreoretinal traction, particularly in patients with large breaks. Intraoperatively, a 360° scleral indentation should be performed to identify all breaks and meticulous peripheral vitrectomy is required. A complete fluid-air exchange, definite laser and cryopexy retinopexy also account for the anatomic success. Furthermore, the sclerotomy should be securely closed to avoid hypotony immediately after surgery. Finally, education on postoperative refraining from vigorous exercise for 1 month and regular follow-ups are also critical.

This study should be regarded as an initial exploration of the prevalence and predictors of retinal breaks reopening in primary RRDs managed by PPV with air tamponade. However, several limitations have hindered the interpretation of our findings. First, this study was retrospective in nature. Patient-reported outcomes may have been subjected to recall bias. The exclusion of patients lost to follow-up 12 months after surgery can also underestimate recurrence rate. Nevertheless, the inclusion of a large and heterogeneous patient sample, the use of rigorous statistical methodology and adjustment of various interactions helped to eliminate bias and made our results more convincing.