Highly myopic macular holes are still a challenge for vitreoretinal surgeons, and various modified surgical techniques based on the ‘scaffold theory’ have been proposed [17, 18, 25, 29]. In 2016, Grewal introduced the ART approach for refractory MHs, suggesting that this flap might work either as scaffold or as macular plug to help sealing the hole [21], reaching anatomical closure rates of around 90% in a successive multicenter study. Moreover, 52.3% of patients whose MH had closed had an improvement in vision [26]. More recently, a global study gathering 130 cases of ART showed an 89% rate of MH closure, reaching 95% of closure in MHRD. In this report, 43% of eyes had a visual improvement of more than 3 lines, mainly associated with EZ reconstitution [30]. Finally, in the CLOSE study group collecting 1135 eyes from 35 articles, ART showed an 87% closure rate in extra-large MHs (> 1000 μm) [31].

In our research on refractory HMMH, anatomical closure was achieved in 78% of cases. However, median BCVA was not improved at the end of the follow-up period: only one patient gained at least 2 Snellen chart lines. Specifically, median BCVA progressively improved in the first 6 postoperative months, but then re-decreased over time. However, since we focused on eyes with significantly high AXLs (mean 31.63 mm), the progression of myopic atrophy maculopathy (MAM) may have impacted on visual outcomes. Moreover, when correcting the dimension of the holes for the AXL using a previously published formula [15], the mean MLD was 771 μm (range 549–1518), indicating a class of patient with poorer anatomical and functional outcomes. In addition, four cases had HMMH-associated RD at presentation, suggesting a pre-existing significant damage to photoreceptor cells. In contrast with our research, in a cohort of highly myopic eyes with MHRD undergoing ART, Li et al. report better post-operative visual improvement. However, in some cases, visual recovery was hampered due to factors such as atrophy of the RPE and choroid and prolonged RD. Several previous studies suggest that, in these complex cases, a permanent harm to the photoreceptor cells has already taken place, regardless of retina reattachment [1, 32, 33]. Finally, even if PFCL was employed for graft harvesting and positioning over the hole, we didn’t use it as an adjuvant in the early post-operative period in order to optimize graft stability, as recently suggested by Moysidis et al. [30]. Previous research also highlights that SO hinders oxygen transfer between the retina and the anterior chamber after PPV [34], thus reducing oxygen diffusion to the transplant. On the other hand, PFCLs theoretically offer better oxygen diffusion and perfusion, which could be critical for the merging of the retinal graft and improve visual outcomes [30].

Previous research reports that ELM and EZ reconstituted after ART, suggesting an unidentified migratory process that may integrate the flap with the original retina tissue, since the boundaries between the graft and MH edge eventually vanished [26]. These results were successively confirmed in the investigations of Thomas et al. and Parolini et al. [35, 36]. Similarly, a recent report by Lumi et al. highlights that microperimetry was able to show retinal function in the peripheral area of the retinal graft, while multifocal electroretinography (mfERG) showed abnormal function of the central ring and normal function of the second ring [37]. By contrast, in HMMH, we highlighted that the margins of the retinal graft were still visible in 75% of eyes, even years after surgery. Only two eyes showed partial reconstitution of ELM and EZ at OCT, and these of the patients with the best visual prognosis. Our results are in line with those of Li et al., whose study focused on 10 cases of refractory MHRD, in which none of the eyes showed restoration of the ELM or EZ [27]. Wu et al. and Ding et al. propose that the retinal graft contributed to the retinal structure and facilitated the restoration of the outer retinal structure at the margin of the MH [25, 38]. Caporossi et al., employing a subretinal hAM patch in recurrent HMMHs, suggest a possible promotion of outer retinal layers regeneration, namely the ELM and EZ, leading to retinal development and differentiation [20].

On the other hand, it is well known that the nutrition and oxygen demand of photoreceptors are mainly met by the choroid [39]. Recent reports highlight that in MHs undergoing successful ART, angiogenesis and anastomosis were seen in up to 35% of the eyes, likely contributing to the survival of the transplanted retina [40]. Tabandeh shows similar findings even in giant recurrent MHs [41]. However, in eyes with pathologic myopia, choroidal impairment is reported to increase proportionality with AXL, with a critical flexion point around 27.26 mm [42]. Starting from this assumption, ART technique in HMMHs may be hindered by a significant pre-existing choroidal vascular deficit, either in terms of visual outcomes, or in terms of graft fusion with the surrounding retina. In fact, we reported two cases of long-term RPE atrophy behind the graft, suggesting that choroidal impairment may hinder the hole healing process, leading to poor integration of the graft and imbalanced homeostasis of the RPE.

In the multicenter study of Grewal et al., graft displacement was the most common intraoperative and early post-operative complication, either with SO or with gas tamponade [26]. Similarly, we highlighted one case of early graft displacement and refractory HMMH at 1-month follow-up. Wu et al. and Liu et al. suggest that using neurosensory retina and autologous blood together might serve as both a bonding agent to lower the chances of graft dislocation post-surgery and as a possible enhancer to speed up the healing process [23, 25]. Furthermore, previous research advises that, in order to facilitate the growth of glial cells by acting as a bridge and scaffold, the retinal graft size should be larger than the diameter of the MH, since the graft may be subjected to retraction over time [21, 30]. Moreover, in highly myopic eyes, in particular with MHRD, the actual dimension of the HMMH is difficult to assess intraoperatively. Consequently, we decided to keep the retinal graft around 1.5 times larger than the HMMH diameter. This allowed to provide sufficient covering by the graft tissue on the hole surface, consistent with what is suggested by previous research [27]. Nevertheless, we reported a case in which the retinal graft progressively contracted and developed cystoid degeneration, leading to re-opening of the hole. This finding suggests that the retinal graft is not fully integrated even months after surgery and may undergo displacement or delayed contraction.

In their first report, Grewal et al. highlighted late development of post-operative CME in 17% of cases, usually around 4–6 months after surgery [21]. In our case series, 3 patients (33%) developed perifoveal CME-like changes starting at month-3 or month-6 follow-ups. These alterations were refractory to dexamethasone therapy and were still visible years after surgery. This edema was characterized by small cysts, predominantly localized in the INL, showing similarities to the microcystic macular edema (MME), previously described in optic neuropathies but recently associated with diabetic retinopathy, epiretinal membranes and ILM peeling. This kind of edema seems to develop as a result of Müller cell dysfunction associated with ganglion cell damage, resulting in poor visual prognosis [43]. Our results are consistent with more recent research, in which post-operative inner retinal edema appeared in up to 44% of cases, being part of graft remodeling and generally not impacting visual functionality [30, 44,45,46]. The failure of RPE pumping system, the breakdown of the blood-retinal barrier, extracellular fluid accumulation, and intraocular inflammation have all been reported to contribute to post-operative CME development [44]. Moreover, in highly myopic eyes, we hypothesize that the presence of the retinal graft may progressively induce further dysfunction of perifoveal Müller cells, leading to the appearance of chronic intraretinal cysts and suggesting that the retinal graft alters the homeostasis of surrounding retinal tissue even months after surgery. In parallel, we observed an increase of graft thickness during the first 6 months post-operatively, followed by a reduction and stabilization of CMT 12 months after surgery. We suggest that the same mechanism giving light to peri-foveal edema, particularly relying on graft remodeling and intraocular inflammation, leads to the swelling of the graft in the first post-operative months, rather than turning back to initial values over time.

Our study has several limitations. First, the study was retrospective and lacked a control group. Second, the sample size was small and included patients with and without MHRD at baseline, thus including patients undergoing either gas or SO tamponade. Nevertheless, patients were followed for at least 24 months and SO had been removed in all cases within the sixth post-operative month. Moreover, the follow-up period was variable among patients and this may have influenced visual outcomes and development of post-operative complications. Finally, other data such as the extent of retinal detachment, the size of the HMMH and the chorioretinal atrophy, might have affected anatomic and visual prognosis, highlighting the need of further research.

In conclusion, we report the results of ART in a small cohort of refractory HMMHs followed for at least two years, highlighting that, even if anatomical outcomes were acceptable in terms of closure rate, visual acuity didn’t show a significant improvement. Moreover, myopic patients showed several adverse effects in the post-operative period, such as HMMH re-opening and microcystic edema like-changes, while only two cases showed effective merging of the graft with the surrounding retina. Overall, in this specific subclass of patients, retinal graft offers excessively variable results, suggesting that an ideal technique has yet to be discovered.