Introduction

Pterygium is a common centripetal growth of subepithelial fibrovascular membrane that may invade the cornea, presumed to originate from altered limbal stem cells. Even though various proteinases and multiple angiogenesis factors have been linked to this abnormal epithelial and vascular growth, the exact cascade leading to the resulted pterygium remains unclear. Exposure to ultraviolet B radiation has been shown to be a major risk factor, shedding light on the high-geographical variance reported for the incidence and prevalence of this condition [1,2,3,4].

Treatment is considered in cases of either intrusion of the membrane into the visual axis, induced astigmatism, or persistent ocular irritation. Treatment options for those cases consist primarily of surgical removal of the membrane followed by different methods for future recurrence prevention, since a sole excision leads to an unacceptable regrowth rate of 24–89% of cases. The use of limbal–conjunctival autograft (LCA) with or without the application of anti-metabolite substances, such as mitomycin C or 5-Fluorouracil (5FU), had been shown to reduce recurrence rate to approximately 5%–15% of cases [5,6,7]. Since the LCA is commonly harvested as a free flap, once sutured or glued over the bare sclera the autograft is avascularized in its nature, receiving its nutrients exclusively by diffusion. Therefore, a proper process of graft revascularization is an important indicator for graft health [8,9,10,11].

Evaluating graft reperfusion can be done by using fluorescein angiography (FA) or indocyanine green angiograph (ICG) [12], but the invasive nature of those imaging modalities limits its use on a regular basis. Optical coherence tomography angiography (OCTA) has emerged as a noninvasive vascular imaging modality in the past decade. Its software compares decorrelation between consecutive OCT scans, using that data to delineate blood vessels. Technically it is done by comparing phase speckle contrast, changes in intensity, or a variation of the full OCT signal [13, 14].

Although primarily designed and used for the purpose of retinal imaging, adjustments in the technique can provide similar vascular imaging of the anterior segment [15]. Although not optimised yet, in recent years, new data regarding the utility of this technology for the evaluation of the normal and pathologic states of the anterior segment vasculature is being gathered [16,17,18,19,20,21,22,23,24,25,26].

A recent case series of 10 patients published by Liu et al. [27], demonstrated the use of OCTA for monitoring the revascularization rate of a femtosecond laser-assisted conjunctival autograft transplant, following pterygium excision. That study focused on vessel regrowth density and its relation to the graft thickness during the mid to late postoperative period.

Data regarding the use of OCTA as a postoperative surveillance tool for estimating early revascularization of LCA autograft is still lacking from the literature. Therefore, the aim of our current study is to investigate the normal conjunctival angiogenesis patterns and tissue healing processes during the early postoperative period, utilising OCTA to perform a qualitative and quantitative measurements.

Methods

Patients

This prospective study included seven patients undergoing pterygium excision with LCA transplantation at Rabin Medical Centre between July and September of 2019. It was approved by the Institutional Review Board and all procedures adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all subjects prior to obtaining the scans and surgery. All surgeries were performed by a single surgeon using the same technique, as described below.

Perioperative and surgical course

The preoperative evaluation included comprehensive biomicroscopic ocular examination, colour photos (Nikon digital camera, model: Naos; attached to slit lamp microscope, HR-elite-mega digital vision, CSO, Scandicci-Firenze, ITALY) and OCTA scans of the anterior segment. All surgeries were performed under local anaesthesia, removing the pterygiums head from the cornea and dissecting a smooth tissue plane towards the limbus, using a flat blade. The remaining pterygium and subconjunctival tenons tissue were then dissected, leaving an area of bare sclera, with its dimensions measured using a calliper. A light cautery was applied to the sclera for haemostasis when deemed necessary. A thin LCA, measured as 1 mm larger than the defect, was harvested from the superior bulbar conjunctiva. At last, the autograft was placed and secured to the surrounding conjunctiva using an 8–0 polyglactin (Vicryl) sutures. The postoperative regimen consisted of topical dexamethasone and ofloxacin four times a day for 1 week, followed by a gradient tapering down over a course of 4 weeks thereafter.

For evaluation of the LCA revascularization post transplantation, patients were imaged on postoperative days (POD) 1, 3, 7 and 30. During each follow-up visit, an OCTA scan of the LCA zone was acquired. For the purpose of orientation, an additional anterior segment colour photo was carried out at each time point.

OCTA scanning

OCTA scanning was acquired using the AngioVue OCTA system (RTVue- XR Avanti; Optovue, Inc., Fremont, California, USA). This instrument has a central wavelength of 840 nm, a band width of 22 μm, a scan depth of 2.3 mm, and an axial and transverse resolutions of 5 μm and 15 μm, respectively. The axial scan rate is 70,000 A-scans per second. OCTA volume scans of the anterior segment measuring 304 × 304 B-scans were acquired in 2.9 s. A drop of Oxybuprocaine 0.4% was instilled into the imaged eye in order to prevent inadvertent eye movement due to corneal dryness during scan acquisition. The contralateral eye was occluded to prevent physiological diplopia, with patients instructed to fixate on a target on their ipsilateral temporal visual field (nasal visual field in cases of temporal pterygium), allowing the scanning zone to be as flat as possible in relation to the source of light. Eye tracking and autofocus functions were deactivated, with manual adjustments of Z motor and focus parameters made. The device scanning protocol of 3.0 × 3.0 AngioRetina was used. Each set of scans acquisition took approximately 4 s, repeated 3–4 times to minimise motion artefacts. The obtained images were further exported from the SLO bar of the software purposely avoiding additional image processing to represent a 'real world' data setting for qualitative assessment.

Qualitative analysis

OCTA scans and colour photos were assessed by two observers in order to find common revascularization growth patterns in a descriptive manner. This qualitative assessment aims at avoiding missing objects of interest in the scans overlooked in a pattern quantitative assessment.

We defined revascularization, when a flow signal was detected as a blood vessel appearing structure in the region of the graft. Satisfactory vascularisation was determined when revascularization was distributed throughout all areas of the LCA.

Quantitative analysis

Vessel density

Images were exported from the system as a portable network graphics image file. Horizontal line artefacts and speckle noise were removed using a program written in MATLAB, designed specifically for this objective, and similar to a method that was described previously [28]. Briefly, a vertical profile of the image was acquired. To clean the noise from the vertical profile a mean filter was used. The profile was then smoothed using a Savitzky–Golay filter. Original lines were replaced with median filters lines. Vessel was enhanced using thresholds (transformation into binary image). The region of interest (the graft) was identified, and a vessel density signal percentage was calculated for that region (Supplemental Fig. 1).

Graft thickness

The graft thickness was measured at each postoperative visit by an independent grader. Analysing the different B-scan passing through the graft, the grader identified the thickest zone. At this region, the thickness was measured using the Built-in calliper of the device (Supplemental Fig. 1), from a point at the exterior border of the conjunctiva of the graft to a point estimated by the grader to be the interior border of the graft.

Results

Patients’ characteristics

A total of 7 eyes of 7 patients (3 males) with an average age of 63.86 (SD 11.32) undergoing pterygium extraction with LCA were included in this case series. Basic characteristics of the patients are presented in Supplemental Table 1. Five eyes (71%) had nasal pterygia. Three patients attended all follow-up scheduled visits, 3 others missed 1 visit and 1 patient, who was not included in the quantitative analysis, missed 2 visits.

Descriptive analysis

On the first day post-surgery, either minimal or no flow signal was shown at the location of the LCA. Regrowth of blood vessels into the graft was detected on the OCTA scans on postoperative day 3–7, with the formation of a nonorganised appearing vessels. Blood vessels were seen growing in a centrifugal pattern towards the surrounding conjunctiva. Since a 'no signal' gap was seen encircling the central part of the graft, nourishment of these vessels seems to originate from the underlying episcleral bed. Upon 1 month after the operation, all grafts achieved satisfactory vascularisation of the entire graft area.

Two cases were specifically chosen for a more detailed description below. They were selected due to their adherence to all postoperative scheduled visits as well as their different healing process time, representing two different normal healing process rates. Their preoperative colour photos and OCTA images are presented in Supplemental Fig. 2.

Case 1

A 75-year-old female referred to our department for pterygium removal. Her initial slit lamp examination revealed a delicate appearing nasal pterygium with extensive growth over the cornea. Surgical removal of her pterygium with LCA transplantation was carried out with no complications. Her longitudinal follow-up photos are presented in Fig. 1. Observing her colour photos on POD 1 and POD3 time points, the LCA could be easily located using the sutures line, but haemorrhages under the LCA limit the view of either the graft or the episcleral circulation. OCTA images showed a flow signal located at a perilimbal zone as early as on POD1. Those vessels demonstrated a centrifugal growth pattern on OCTA images on POD3 and POD7, with no clear connection to the surrounding conjunctival circulation system seen. On POD30 revascularization was almost complete, with relatively minor areas of absent flow signal.

Fig. 1: Colour photos and their related OCTA scans of patient 1 during POD 1, 3, 7 and 30.
figure 1

A On POD1, a flow signal is already detected near the limbus, pointing to an episcleral origin. B, C On POD3 and POD7, the vessels demonstrate a centrifugal growth, with no clear connection between those vessels and the surrounding. D On POD30 blood supply to the autograft is of episcleral and conjunctival origin, with minor no signal areas indicated by the arrows.

Full size image

Case 2

A 57-year-old female with persistent ocular irritation was referred to our department for pterygium removal. Her preoperative evaluation demonstrated a fleshy pterygium extending 2 mm over the cornea on the temporal side of her left eye. Her longitudinal follow-up images post Pterygium removal with LCA implantation are presented in Fig. 2. On POD1, a thickened and oedematous LCA was accompanied by diffuse haemorrhages under the graft. The same clinical picture was also evident on POD3. At the same time point, a gap between the graft and the adjacent temporal conjunctiva was seen. No intervention was carried out and the gap was left to heal by primary intention. OCTA scans of the LCA showed no flow signal on both POD1 and POD3 visits. On POD7, a vascular network growing from the underlying episcleral vascularisation was seen. Gaps between those vessels and conjunctival vessels around the graft are clearly seen (Fig. 2). One month after the surgery, an almost complete revascularization was observed, with only limited areas of absent flow signal still evident. Cross-sectional OCT B-scans at different time points are presented in Fig. 3, showing revascularization progresses as the LCA becomes thinner and less oedematous, with disappearance of intra-conjunctival cysts.

Fig. 2: Colour photos and their related OCTA scans of patient 2 during POD 1, 3, 7 and 30.
figure 2

During follow-up, a small gap between the autograft and surrounding conjunctiva was found, left for healing by primary intention (Blue arrows). A–C The flow void detected on POD1 and POD3 is followed by a network growth of blood vessel at the centre of the autograft on POD7. A clear connection between this network and the marginal conjunctival vasculature is still not seen. D A centrifugal pattern of vessel growth is evident on POD30, with small areas of absent signal, indicated by green arrows.

Full size image
Fig. 3: OCT B scans of patient 2 during POD 1, 3, 7 and 30, with the corresponding en-face images to the left of each image, indicating the location of the OCT scan.
figure 3

A–D Intra-conjunctival cysts are evident in the autograft during POD1 and POD3. As the revascularization process advances during POD7 and POD30, no cysts are further observed, with graft thickness gradually decreasing.

Full size image

Quantitative analysis

Table 1 shows the percentage of vessel density signal measured for each visit. Vessel density signal is very low on POD 1 (Mean 7.1 ± 3.3%), rising in the next two days (8.7 ± 3.6%) and furthermore on POD7 and POD30 (14.3% ± 4.1% and 21.6% ± 2.2%, respectively). Grafts demonstrating higher initial vessel density signal on POD1 maintained that superiority on POD3 and somewhat on POD7 as well, but not on POD30 (Supplemental Graph 1).

Table 1 Vessel density (%).
Full size table

Table 2 shows graft thickness measurements (in μm) for each visit, affected mainly by oedema of the graft. It appears that thickness rises during the early period from POD1 (Mean 611 ± 120 μm) to POD3 (695 ± 84 μm). (Oedema is still substantial on POD7 (639 ± 96 μm), along with progression of vessel density flow signal, with substantial decrease in thickness measured on POD30 (300 ± 108 μm) (Supplemental Graph 2).

Table 2 Graft thickness (μm).
Full size table

Integrating the data from the two graphs, it is noticeable that grafts with lower vessel density on POD1 and POD3 (patients 2, 3 and 6) were more oedematous compared to the other grafts.

Discussion

Anterior segment OCTA is emerging as a valuable tool for gathering functional information regarding the ocular vasculature flow. In this current work, we assessed the revascularization of LCA in a cohort of patients undergoing pterygium excision, using an OCTA scanning system. Patterns of blood vessel growth were sought, tracking OCTA scans, and colour photos at the early postoperative period.

Free conjunctival grafting involves the transfer of autologous conjunctival cells while preserving the anatomic orientation but not the original blood supply. A similar process of free tissue transfer can be witnessed in free skin grafts, enabling us to draw an analogy to the three phases seen during skin graft recovery [29]. In the first few days post transplantation, named serum imbibition phase, the graft obtains oxygen and nutrients by diffusion through the plasma. In the first hours, passive absorption of serum from the wound bed causes oedema, which resolves when revascularization becomes functional. Back in 1969, Converse et al. [30] showed that skin grafts gain up to 40% of their initial weight within the first 24 h, and that this gain is reduced to 5% at 1 week post grafting. During the second phase, occurring as early as 24–48 h post transplantation, revascularization occurs. Patterns of revascularization are mainly described as vessel ingrowth from the bed of the wound rather than wound margins, supported by the lack of evidence of an increase in blood vessels density or distribution in grafts margins [29, 31]. There is still an ongoing controversy whether the revascularization process stems from neovascularization, where new blood vessels originate from the recipient site into the skin graft, or from Inosculation, where an anastomosis between the host and the graft own vessels occurs [31, 32]. It is possible that a combined process is involved, with integration of the two patterns of vascular formation. During the last phase, maturation of reforming blood vessels takes place, when the graft and the host had completely integrated, with remodelling and contraction of the tissue occurring.

While there are clear differences between these tissues, our results demonstrate that the skin grafting phases described above could apply to LCA reception by the bare sclera as well. On the first postoperative day, OCTA scans showed no to minimal flow signal. At this time point, the graft was significantly thicker, as demonstrated in both colour photos and cross-sectional OCT measurements. This observation points to a similar serum imbibition period after LCA transplantation.

Starting from the third POD to the seventh POD, a centrifugal expansion of the flow signal was detected. The observation of flow signal gaps in the interface of the graft and the host conjunctiva, supports the notion that revascularization originates from the underlying episclera rather than the grafts borders, resembling the aforementioned second phase of skin grafts. This observation might emphasise the importance of measures undertaken to preserve the episcleral vascularisation integrity during surgery, by limiting tissue cauterisation for example.

Another observation relates to the effect of revascularization on the oedematous status of the graft during the early postoperative period. Generally, oedema was increasing during the first postoperative week along with progression of revascularization, and eventually decreased substantially by POD30. In addition, we noticed that grafts with lower vessel density signal (that reflects lower vascularisation) were more oedematous during the first postoperative week. That observation may reflect the process described during the early phase after skin graft transplantation, where the oedema continues and resolves as soon as revascularization becomes functional. In some cases, these early postoperative factors can reflect clinically significant ischemia that may play a role in the pathogenesis of pterygium recurrence [33]. Furthermore, studies can elaborate more on that connection as well as assess whether existing treatments (e.g., Topical Steroids or subconjunctival anti-VEGF [33]) affects these patterns and recurrence rates. We believe additional data in that field may utilise OCTA as a clinical tool in the postoperative period, for identifying patients at a higher risk for graft failure and pterygium recurrence and treat them early.

Previous studies using ICG in anterior segment angiography scans coincide with the results of our current study. Chan et al. [34] demonstrated graft reperfusion 1–2 weeks following conjunctival autografting, with obvious gaps between the graft and the host conjunctiva, in 9 patients post pterygium excision. This finding led them to conclude that the source of reperfusion lies in the episcleral bed, resulting in a fully perfused with mild leakage grafts at 2 months post transplantation. Aydin et al. [35] evaluated the reperfusion of LCA 1, 7 and 30 days after transplantation, using FA and ICG angiography in 12 eyes of 11 patients. On POD1, ICG angiography showed that all 12 grafts were diffusely hypofluorescent, with multiple hyperfluorescent foci appearing at the graft margin in the late phase. By POD7, fluorescent signal increased and vessels growing into the graft were evident, originating from the episcleral vascular bed. On day 30, grafts were isofluorescent in 10 eyes. By this stage, perfusion of the graft was completed. In another publication by Kim et al. [36], ICG findings were consistent with episcleral vasculature early reperfusion theory, with blood vessels remodelling proceeding for 3 months postoperatively.

As stated before, the utilisation of ICG angiography for anterior segment imaging is limited by its invasive nature, while OCTA imaging is a noninvasive, fast procedure which provides segmentation as well. A recent published study [37] showed OCTA to be superior over ICGA for corneal neovascularization assessment in an animal model. It delineates corneal vessels and provides inter-observer measurement better but may overestimate vessel density.

In a recent study by Liu et al. [27] ten patients were followed by OCTA at intervals of 1 week, 1 month and 3 months after pterygium excision and femtosecond laser-assisted conjunctival autograft transplantation. The vessels density at three different depths was evaluated and quantified, enabling them to calculate the vessel regrowth rate. Due to their study design, patients were not followed or imaged on the first and third days after surgery, not allowing for exploration of patterns of vessel regrowth and their origins during the first postoperative week. Nevertheless, they found a significant and strong negative correlation between vessel regrowth density and the changes of conjunctival autograft thickness, similar to our findings. They concluded that graft reperfusion plays an essential role in the resolution of tissue swelling after surgery.

There are several limitations to our study, that warrant consideration. The major ones are the small number of patients included and the lack of adherence to the scheduled visits by some of the patients. Another limitation lies in the technology itself [27]—the current OCTA system is not optimised for anterior segment imaging, lacking eye tracking and autofocus capabilities, with efforts undertaken to overcome fixation losses and motion artefacts. The lack of an option to scan the same exact spot, as easily done in the field of retinal imaging, limits longitudinal follow-up of the exact same point. In addition, the automatic segmentation system is based on the retinal structures, and thus is not suitable for the anterior segment. Manual segmentation of each B scan is possible but its use in the daily clinic work is very cumbersome and somewhat limited by the machine software. Therefore, in this current study, we used whole thickness signal images, easily repeatable under daily clinic conditions. Another limitation lies in the comparison of spectral domain OCTA which we have used, to Swept source OCTA which is an emerging OCTA technology. It may have possible advantage due to faster scanning speed that allows for denser scan patterns and larger scan areas. Another advantage of swept source OCTA is the use of a longer wavelength resulting in enhanced light penetration through deeper tissues and better imaging of deeper blood vessels [38].

In conclusion, OCTA imaging may have an important role in the evaluation of graft health and reception during the early postoperative period. Future studies could help shed more light on the effect of different phases of surgery on the reperfusion rate. Such data provided by OCTA studies may help surgeons further understand the implications of their actions during different surgical steps on graft integrity, such as the choice between fibrin glue or sutures, the amount of appropriate vascular cauterisation, the initial graft thickness, manual versus laser-assisted graft preparation, and postoperative care management. It would be also interesting to investigate whether an association exists between some specific early graft revascularization patterns and pterygium recurrence.

Summary

What was known before

  • The source of LCA reperfusion lies in the episcleral vessels.

  • Revascularization takes place on the 7th postoperative day.

  • Revascularization is negatively correlated with LCA thickness.

What this study adds

  • Revascularization of LCA occurs between the 3rd and the 7th postoperative day.

  • The healing process after free conjunctival grafting follows the same phases as those known to occur after free skin graft.

  • OCTA imaging may have an important role in the evaluation of graft health and reception during the early postoperative period.

  • The graft is oedematous in the first postoperative week then oedema subsides as revascularization process occurs.