Introduction

Although first described by Gass [1] as a partial thickness macular lesion, the description of  lamellar macular hole (LMH) has been further refined by the advent of optical coherence tomography (OCT). In fact, structural analysis of LMHs revealed that they are generically characterized by a partial thickness foveal defect with avulsion of the inner layers of the retina and a thin residual fovea [2]. Nowadays, two different types of primary LMH are identified: a degenerative LMH and a tractional LMH. Degenerative type is characterized by a slow progressive tissue loss without evidence of tractional component [3]. Distinctive OCT characteristics are the presence of irregular foveal contour (i.e., abnormal, non- linear shape of the foveal pit contour), foveal cavity with undermined edges and the presence of at least one other sign evoking a loss of foveal tissue (i.e. pseudo-operculum, thinning of the foveal at its centre or around) [4]. Moreover, the presence of a lamellar macular hole associated epiretinal proliferation (LHEP) [5], of a foveal bump and of an foveal or parafoveal ellipsoid zone (EZ) defect are frequently associated. By contrast, the presence of a contractile epiretinal membrane (ERM), retinal thickening or wrinkling and the appearance of cystoid spaces in the inner nuclear layer (INL) define OCT characteristics of tractional LMHs [4]. Lastly, the presence of tractional elements in the context of a degenerative LMH define a mixed type LMH. Even though the pathogenesis of both tractional and degenerative types is related to the occurrence of an anomalous posterior vitreous detachment (PVD) as a first event [3, 6,7,8,9,10], recent evidence highlighted the contribution of microvascular blood supply changes to the progression of the degenerative LMH [11, 12]. LMH-like lesions, whose pathogenesis differs from the one of primary LMHs [4], may also be present as a consequence of unroofed cystoid macular oedema, end-stage age related macular degeneration [13], macular telangiectasia [14], solar retinopathy, tamoxifen retinopathy [15] and partial closure of full-thickness macular holes. Degenerative LMHs are characterized by a poorer response to surgical treatment compared to their tractional counterpart [16] and both the amount and the speed of tissue loss within the lesion have been linked to the functional prognosis in patients treated with simple surveillance [11]. Nonetheless, spontaneous closure of LMHs undergoing wait and watch treatment strategy was sparsely described in scientific literature, even though no systematic report of basic characteristics and predisposing factors to this event has been provided yet [17, 18]. The aim of the study is therefore to the describe OCT imaging characteristics of a cohort of patients showing spontaneously closing degenerative or mixed type LMH and to compare them to the ones of a sex, age and follow-up matched group showing stable lesions.

Materials and methods

Retrospective screening of patients diagnosed with degenerative or mixed type LMH between January 2011 and April 2022 was conducted in 3 different specialized retina centres: Department of Ophthalmology of Hopital Fondation Adolphe de Rothschild, Department of Ophthalmology of Humanitas Gavazzeni-Castelli and Department of Ophthalmology of Hopital Intercommunal Paris Creteil. The study was conducted in accordance with the Declaration of Helsinki and with French and Italian legislation and was approved by local Ethics Committee. Informed consent was obtained by all patients. Cases showing spontaneous anatomical closure at last optical coherence tomography (OCT) examination were selected to take part to the SC (spontaneous closure) group. Spontaneous closure was defined as the presence of a progressive continuous process of reversal of tissue loss resulting in relevant restoration of foveal anatomy in patients with previously documented LMH. By consequence, all included patients featured at least 2 follow up mode high quality fovea crossing OCT B scans documenting the presence and the subsequent closure of the LMH. All included images were acquired with Spectralis HRA + OCT (Heidelberg Eye Explorer, Heidelberg Engineering, Heidelberg, Germany). An image quality score > 15 was deemed as necessary for inclusion [19]. An equal number of subjects matched for age, sex and follow up duration were randomly selected among patients with anatomically stable degenerative and mixed type LMHs to take part to the ST (stable) group. Exclusion criteria for both groups were ocular surgery of any type during the follow up period, concomitant presence of ocular inflammatory diseases, history of ocular trauma, history of full thickness macular hole, exudative age-related macular degeneration, macular telangiectasia and present or past tamoxifen therapy. The time of the last available acquisition before the beginning of the regression process was defined as “time 1” (T1) while the time of first detection of spontaneous closure was defined as “time 2” (T2). The interval between the two was defined as follow up time. All images were evaluated by two graders. In case of disagreement, a third independent senior grader expressed the final decision.

Age, sex, duration of follow up and best corrected visual acuity (BCVA) at T1 and T2 were collected for each participant. BCVA decimal values were converted to logMAR values. The presence or absence of complete posterior vitreous detachment (PVD), lamellar hole associated epiretinal proliferation (LHEP) and ellipsoid zone (EZ) interruption both at T1 and T2 examination were assessed based on unprocessed OCT B scan images. EZ interruption in the fovea or parafovea (3 mm diameter circular zone centred on the fovea) was defined as focal or extensive hyporeflective disruption in the continuous hyperreflective line corresponding to IS/OS junction lying between the interdigitations of the OS of photoreceptors and RPE and external limiting membrane [20]. Foveal and parafoveal inner retina was searched for the presence of hyperreflective foci (HF). HF were defined as intraretinal spots with a reflectance higher than that of the retinal nerve fibre layer (RNFL). The inner retina, whose boundaries were defined as the inner limiting membrane (ILM) (inner boundary) and the inner plexiform layer (IPL) (outer boundary), was segmented and divided in equal regions with a surface of 0.05 mm2. Mean HF density was calculated for each group at both T1 and T2. OCT images were then processed using the mean reflectance of the RNFL from each image as a thresholding limit. HF presence and density was then reassessed by the graders in images from both T1 and T2.

The presence of a hyperreflective inner border (HIB) lying on the margins of the loss of substance was assessed in unprocessed images at T1 and was defined by > 33% of the cavity margin showing relative hyperreflectivity. Images were then processed using ImageJ software (ImageJ, NIH, Bethesda, MD). Processing consisted in binarization of the image and application of local adaptive thresholding based on mean values with a kernel radius of 5. The presence of HIB was subsequently reassessed on processed images.

Lastly, the presence of a linear hyperreflectivity in the outer plexiform layer (LHOP) in foveal and/or parafoveal region was registered at both T1 and T2 in unprocessed images. ImageJ software (ImageJ, NIH, Bethesda, MD) was used to binarize images and apply a mean based local adaptive thresholding based on local mean (radius = 5). Processed images were subsequently re-evaluated for the presence of LHOP.

Statistical analysis

Statistical analysis and sample size calculation were conducted with SPSS software v26.0 (IBM, Armonk, New York, USA). A p-value < 0.05 was considered as statistically significant. Normality of the distribution for quantitative variables was assessed with Shapiro Wilk test. Univariate analysis was conducted using two-tailed t test for independent samples for normally distributed variables and Mann-Whitney test for non-normally distributed variables. Paired T-test was used to analyse changes in BCVA within the same group from T1 to T2. Qualitative variables were compared using Chi square test and Fisher’s exact test when appropriate. For qualitative variables showing borderline (p < 0.09) or full significance (p < 0.05), odds ratio (OR), sensitivity, specificity, and likelihood ratio (LR) were also calculated. For OR Baptista-Pike method was used. Sensitivity, specificity and LR were assessed with Wilson-Brown method.

Results

Of 187 degenerative LMHs screened, 11 eyes of 11 patients showed spontaneously closing degenerative LMH (SC group). Mean age was 69.2 ± 7.1 years and male sex had a prevalence of 54.5% (6/11) (see Table 1). The control group (ST group) showed no differences in age (p = 0.96) and gender (p > 0.99) composition and in duration of the follow up (p = 0.81). BCVA at T1 was 0.151 ± 0.017 logMAR in SC group and 0.155 ± 0.041 logMAR in ST group, which was not statistically significant (p = 0.79). BCVA at T2 significantly differed between groups (p = 0.004). A borderline significance in prevalence of LHEP at T2 was detected between groups (p = 0.085), corresponding to a sensitivity of 63.6% (CI 35.4–84.8%), a specificity of 81.8% (CI 52.3–86.7%) and a LR = 3.50 for the detection of SC degenerative LMHs. The prevalence of EZ interruption was not statistically significantly different between groups both at T1 (p = 0.99) and T2 (p = 0.362). The presence of a HIB in unprocessed images at T1 was characterized by a borderline difference between groups (p = 0.086, sensitivity 72.7% (CI 42.3–90.3%), specificity 81.8% (CI 52.3–86.7%), LR = 2.667). Processed OCT images revealed the presence of HIB in 10/11 (90.9%) patients in SC group and 3/11 (27.3%) patients in the ST group, resulting in a statistically significantly higher prevalence in SC group (p = 0.007) with a sensitivity of 90.9% (CI 62.3–99.5%), a specificity of 72.7% (CI 43.4–90.3%) and a LR = 3.33 (see Fig. 1). Notably, image processing increased sensitivity for HIB detection in SC group. The presence of LHOP in unprocessed images was detected in 4/11 patients in SC group and 1/11 patients in ST group at T1 and in 3/11 patients in SC group and 1/11 patients in ST group at T2. Processed images revealed the presence of LHOP in 8/11 patients in SC group and 2/11 patients in ST group at T1 (p = 0.03), displaying a sensitivity of 81.8% (CI 52.3–96.8%), a specificity of 72.7% (CI 43.4–90.3%) and a LR = 3.00. The presence of LHOP in processed images didn’t differ between groups at T2 (p = 0.183). HF foci density wasn’t significantly different between groups at both unprocessed and processed image at T1 and T2. Detailed description of the population is provided in Table 1.

Table 1 Comparison of spontaneously closing degenerative LMHs (SC group) with stable degenerative LMHs (ST group) on anamnestic, imaging and functional variables of interest.
Full size table
Fig. 1: Follow up mode OCT B scan sections showing spontaneous closure of a degenerative LMH during a 4 year follow up.
figure 1

The right column displays unprocessed OCT images while the left column shows processed OCT images. The red arrow highlights a zone of HIB in the area most interested by the process. In this case, image processing evidenced an HIB that was almost imperceptible in unprocessed image, showing an example of how the processing technique increased sensitivity to the sign. First row: the last acquisition before the beginning of LMH closure process (T1); Second row: intermediate acquisition (2 years follow up); Third row: first acquisition after LMH closure (T2). HIB Hyperreflective inner border, LMH Lamellar macular hole; OCT Optical coherence tomography.

Full size image

Two hundred (200) eyes diagnosed with mixed type LMH were screened. Ten [10] patients showing spontaneously closing mixed type LMH (SC group) with a mean age of 67.4 ± 7.9 years and a 50.0% prevalence of male sex were selected along with an equal number of patients showing stable mixed type LMH (ST group). The two groups showed no significant differences in age (p = 0.98), gender (p > 0.99) and follow up time (p = 0.98). No significant differences in BCVA between groups at both T1 and T2, even though SC group experienced a statistically significant improvement in BCVA during the follow up time passing from 0.104 ± 0.016 logMAR to 0.127 ± 0.033 logMAR (p = 0.025) (see Table 2). LHEP was detected in 2/10 patients in SC group and 2/10 patients in ST group at T1. One patient in the SC group developed LHEP at T2 (3/10 patients) but the difference was still not statistically significant (p > 0.99). The prevalence of EZ interruption wasn’t significantly different at both and T2 T1 (p = 0.629 and p = 0.99). HIB was detected in unprocessed images in 5/10 patients in SC group and 2/10 patients in ST group (p = 0.349) while processed images showed HIB in 7/10 patients in SC group and 3/10 patients in ST group (p = 0.178). Without image processing, LHOP at T1 was present in 4/10 patients in SC group and 0/10 patients in ST group (p = 0.086), showing a specificity of 72.2%. With image processing, at T1 8/10 patients (80%) showed LHOP as opposed to only 1/10 patients in the ST group (p = 0.005). These differences resulted in a sensitivity of 80.0% (CI 49.0–96.5%), a specificity of 90.0% (CI 59.0 –99.5%) and a LR = 8.00. At T2, unprocessed images showed LHOP in 3/11 patients in SC group and 0/11 patients in ST group (p = 0.210). Processed images at T2 showed LHOP in 6/10 patients in SC group compared to 1/10 patients in ST group (p = 0.057). An example of a spontaneously closing mixed-type LMH featuring LHOP is provided in Fig. 2.

Table 2 Comparison of spontaneously closing mixed type LMHs (SC group) with stable mixed type LMHs (ST group) on anamnestic, imaging and functional variables of interest.
Full size table
Fig. 2: Follow up mode OCT B scan sections showing spontaneous closure of a mixed-type LMH during a 4 year follow up.
figure 2

The left column displays unprocessed OCT images; the white circles highlight a zone of LHOP mostly visible at 1-year follow up acquisition. The left column shows processed OCT images; LHOP is highlighted with a red arrow. First row: the last acquisition before the beginning of LMH closure process (T1); Second row: intermediate acquisition (1 years follow up); Third row: intermediate acquisition (3 years follow up); Forth row: first acquisition after LMH closure (T2). LHOP Linear hyperreflectivity in the outer plexiform layer, LMH Lamellar macular hole, OCT Optical coherence tomography.

Full size image

Discussion

The optimal management of patients diagnosed with degenerative LMH is controversial due to the relative anatomical and functional stability of these lesions and to unpredictable response to surgical treatment [21]. In this context, the notion of the possibility of a spontaneous closure assumes particular relevance. We describe a series of 11 degenerative LMHs and 10 mixed type LMHs encountering spontaneous closure. The median duration of the process was 4 years (range 2.1–8.3 years). Considering the number of screened patients, the esteemed prevalence was 1 in 20 (10/200, 5.0%) LMHs for mixed type and 1 in 17 (11/187, 6.0%) LMHs for degenerative type. Nevertheless, given the retrospective nature of the study, no precise inference on timing and incidence can be provided. We also performed a comparison with an equal subgroup of stable LMHs. As concerns degenerative LMHs, spontaneously closing ones showed a higher prevalence of HIB at both processed and unprocessed images just before the beginning of the closing process (T1). Moreover, a significantly higher prevalence of LHOP was detected in processed images at T1. As concerns mixed LMHs subtype, the difference in LHOP prevalence between spontaneously closing and stable group was even more evident: unprocessed images showed a borderline significance at T1 between the two and processed images showed a significant difference at both T1 and T2. By contrast, mixed type LMHs didn’t show significant differences in terms of HIB prevalence both at processed and unprocessed images. Neither degenerative type nor mixed type spontaneously closing LMHs showed differences in terms of HF density compared to their stable counterpart. Lastly, at the end of the closing process, LHEP was present in 7/11 patients showing spontaneous closure compared to 2/11 stable patients, a difference which was borderline to statistical significance. Interestingly, two patients developed LHEP during the follow up period and all the 5 patients already showing LHEP at T1 experienced thickening of the proliferation. This is consistent with the finding reported from Chehaibou et al. [18], that described the case of two LMHs with progressive thickening of the LHEP encountering spontaneous closure. Similarly, Compera et al. [22]. highlighted how the development of a LHEP seems to be associated with changes in cavity shape of degenerative LMHs. All the described differences between spontaneously closing and stable LMHs seem to point out at a glial activation-mediated process. We believe that LHOP, which we describe for the first time in literature, might in fact be the signal of an activation process starting in the OPL in which microglia-Muller cells interactions play a pivotal role. Retinal microglia is prevalently distributed in the inner and outer plexiform layer and is fundamental to tissue homeostasis, immunoregulation, and repair [23]. Its activation has been ascribed both beneficial and deleterious consequences in response to pathological stimuli [24]. Ramified parenchymal microglia in the adult retina originates from the differentiation of CD45 and major histocompatibility complex (MHC) class I- and II-positive precursors appearing in the retina prior to vascularization while a second category of microglial precursors, which do express specific macrophage markers, migrate into the retina along with vascular precursors and appear around blood vessels in the adult retina [23]. In vitro studies showed that Müller cells exposed to microglial activation exhibit marked alterations in cell morphology and gene expression that are adaptive in nature and differ from those seen in chronic gliosis [9]. These changes include upregulation of growth factors such as glial cell-derived neurotrophic factor (GDNF) and leukaemia inhibitory factor (LIF) and upregulated chemokine and adhesion protein expression, which allow Müller cells to attract and adhere to microglial cells thus promoting migration. Moreover, this process is perpetrated by the fact that modified Muller cells produce pro-inflammatory factors generating a positive feedback loop interaction [25]. We believe that this process might be at the origin of the phenomenon of spontaneous closure and that both HIB and LHOP might be the macroscopic sign associated to its activation. Whether the trigger to the process is represented by antigenic or endogenous, biological or mechanical stimulation remains unclear. Interestingly, mechanical stress induced by PVD is mainly exerted at the level of INL and OPL. Traction on this layer may induce stimulation and migration of resident microglial cells, giving way to LHOP formation in specific circumstances [26]. In mixed type LMHs, the process might be triggered by specific changes in tangential tractional forces. By contrast, in degenerative LMHs, changes in microvascular blood supply might amplify the process of microglial activation. In fact, previous literature evidenced how microvascular changes are associated to the anatomical and functional progression of degenerative LMHs. Another possible explanation is that a release of epiretinal traction could be responsible for spontaneous closure in the case of mixed type LMHs. As visible in Fig. 2, enface infrared images showing the evolution of the lesion do not show signs of tractional release. Unfortunately, being a retrospective study, OCTA was not available for all patients from degenerative LMH group and so we weren’t able to investigate the likelihood of this hypothesis. Interestingly, even though HF are widely associated with intraretinal inflammation [27, 28], HF density was not found to be significantly different between spontaneously closing and stable LMHs. Similarly, stable and spontaneously closing LMHs didn’t show differences in terms of complete PVD and EZ interruption. Lastly, it should be noted that BCVA at the end of the process significantly differed between spontaneously closing and stable degenerative LMHs and that both degenerative and mixed type LMHs showed a significant increase in BCVA during follow up time. Limitations of the study include the small sample size and the retrospective nature of the study, which impedes accurate assessment of the incidence of spontaneous closure. Moreover, availability of OCTA acquisitions would have allowed investigation of the influence of the microvascular component in the closure process. Future research including prospective analysis of the phenomenon of spontaneous closure, featuring also the evaluation of the role of the described potential triggers, is therefore highly encouraged.

Summary

What was known before

  • Lamellar macular hole are vitreoretinal lesions that occur secondary to anomalous posterior vitreous detachment causing vision loss

  • Evidence has shown that some LMHs show progressive behaviour

  • Cases of spontaneous closure of a LMH have been reported in literature

What this study adds

  • Spontaneously closing LMHs show higher prevalence of hyper reflective inner border (HIB) and linear hyperreflectivity in the outer plexiform layer (LHOP) at the beginning of the closing process. These signs are here described for the first time in literature

  • These signs are best enhanced with image processing methods

  • They might be the consequence of microglial and Muller cells coordinated activation