The cosmetic fillers used in all the 20 enrolled cases were HA gels, which is consistent with reports in the literature that the application of HA was more common than autologous fat16. Of the 20 cases, 5 ones developed PB. However, among the 10 cases of similar ocular complications caused by autologous fat reported in the literature17–24, only 1 patient24 who died showed PB. It can be seen that the possibility of PB caused by HA embolism (5/20) was far greater than that of autologous fat (1/10). We speculate that the possible reason is that HA microspheres are much smaller than autologous fat particles16, so they are more likely to block the terminal branches of the ophthalmic artery (OA), including the ASC, resulting in ASI. The obstruction of the ASC, ranging from mild to severe, represents different degrees of ASI, resulting in CHS, hypotony, and PB, respectively (Fig. 2).
Regarding the definition of hypotony, although the World Glaucoma Association recommended a definition of less than 6.5 mmHg9, the ocular trauma research literature often defined it as less than 8 mmHg7. In the cases of ASC embolization we observed, the IOP below 8 mmHg often showed variety degrees of ASI, and the mildest positive sign was Tyndall effect, so we adopted the definition in the ocular trauma study. The sASI caused by HA mainly occurred in NISS grades 3 and 4, which were characterized by immediate NLP. Surprisingly, the ASC was generally not affected before central retinal artery occlusion (CRAO) occurred. The possible reason is that the central retinal artery (CRA) is a terminal vessel without collateral circulation, and the diameter of the vessel is small, usually less than 0.2mm25; while the ASC usually consists of 7 anterior ciliary arteries (ACAs) and 2 long posterior ciliary arteries (LPCAs), which has abundant anastomoses with the conjunctival and ciliary arteries10,11. Thus, CRAO requires only a small amount of HA microspheres to occur; clinically significant ASI requires a large number of HA microspheres to retrograde into the ASC. This theoretically explained why sASI usually occurred in NISS grades 3 and 4 cases with immediate NLP, with NISS grades 3 and 4 indicating retrograde entry of large amounts of HA microspheres into the OA, secondary occurrence of CRAO, and consequent greater likelihood of sASI.
According to our observation, PB caused by HA embolism mostly (4/5) occurred about half a year after the injury (Table 1). The interval from HA embolism to PB was similar to 0.9 years of infectious endophthalmitis, but significantly shorter than 1.4 and 2.9 years of ocular trauma and uveitis, respectively26. HA embolism-induced hypotony and CHS at 2-year follow-up mostly (3/5) occurred within 1 month after injury, and this pathological state usually remained unchanged for subsequent 2 years. A various recti recession experiment with monkey eyes10 also found that the low IOP usually occurred within 1 month after the surgery. The difference was that the low IOP caused by the recti recession surgery will usually recover spontaneously with the stabilization of the ASC, and generally no PB will occur. It can be seen that the consequences of ASI caused by extensive embolization of ASC by HA were much more serious than those of recti recession surgery.
With regard to the pathophysiological mechanism of PB, we speculated that a large number of HA microspheres retrograde through the OA first entered the extraocular muscular artery (EMA), ACA, LPCA, and CRA, leading to the embolism of these arteries. The ACA is derived from the EMA, which together with the LPCA and the scleral perforating branches of the conjunctival artery forms the ASC10. Subsequently, The CRAO resulted in immediate NLP; occlusion of the EMA resulted in ophthalmoplegia; extensive occlusion of the extraocular muscular-anterior ciliary artery (EMACA) and the LPCA together resulted in severe hypoperfusion of the ASC, ultimately leading to sASI. The co-occurrence of 360-degree circumferential iris atrophy and hypotony within 6 months after extensive occlusion was the most important features of sASI. As a result, sASI meant a very high probability of PB at 6-month follow-up.
In this study, ASI was graded mainly based on the degrees of ciliary process dysfunction, rather than iris fluorescein angiography and anterior segment abnormalities (such as segmental iris atrophy, pupil irregularity, uveitis) adopted in previous article 27. Ciliary process dysfunction was mainly reflected in the changes of IOP, so the grading system we proposed mainly focused on the fluctuations of IOP and whether it was combined with Tyndall effect reflecting the disruption of the blood-ocular barrier. This grading system based on dysfunction had several advantages. First of all, it was convenient, easy to master, and suitable for widespread promotion. Secondly, the grading system did not require special equipment, so it was non-invasive and suitable for assessment and dynamic observation at any time. Last but not least, it could accurately reflect the severity of ASI and the possibility of subsequent PB.
Of course, there are still some deficiencies that deserve attention in this study. First, due to the rarity of such catastrophic consequences, only 10 long-term complications of ASI occurred even after 4 years of enrollment. Therefore, the scientific nature of the grading system still needs to be verified by a large number of clinical practices. In addition, this is a retrospective study, and some patients lack IOP follow-up data at certain time points; some patients lack detailed anterior segment evaluation, such as ultrasound biomicroscopy and gonioscopy, etc. Even so, we still believe that the grading system of ASI mentioned in this paper has a more accurate role in evaluating the long-term ciliary function damage caused by rectus muscles regression surgery or occlusion of ASC, and it is worthy of clinical reference.
Several representative cases are as follows:
Case No. 11
A healthy 35-year-old woman complained of blurred vision, diplopia and orbital pain in her left eye following the injection of HA into her forehead at a private practice on February 13, 2018. Nine days later, she was sent to our hospital. On physical examination, her visual acuity was 20/20 OD and 20/50 OS. IOP (NCT) was 8.7 mmHg OD and 8.9 mmHg OS. Ocular anterior and posterior segments were not detected to be abnormal in the right eye. Exotropia along with medial rectus paralysis, mild ptosis and the positive relative afferent pupillary defect (RAPD) sign were detected in the left eye. The cornea, anterior chamber, iris and lens were normal, but the inferior temporal retina was pale and mildly haemorrhagic in the colour fundus photograph of the left eye (Fig. 4A).
The patient was diagnosed with branch retinal artery occlusion and treated with topical timolol maleate and intravenous alprostadil for 2 weeks and a systemic steroid (dexamethasone 10 mg/d) for 3 days.
Three months later, the blurred vision of the left eye improved greatly, and the inferior temporal pale area in the colour fundus photograph had faded, with the appearance of narrowed branch retinal arteries and pale optic discs (Fig. 4B).
Case No. 13
A 44-year-old woman presented with sudden visual loss of the left eye following receiving the HA injection at a beauty salon on October 12, 2018. She underwent interventional thrombolysis in the nearest hospital the day after the accident, but there was no effect. She came to our hospital for further diagnosis and treatment after the catastrophic consequence.
Routine ocular examinations at the first visit in our hospital included visual acuity of 20/20 OD, 20/200 OS and IOP of 16.2 mmHg OD, 15.1 mmHg OS. Left eyeball movement suggested exotropia in primary eye position and paralysis at upward, medialward, downward gaze. Eyelids inspection showed left swollen upper eyelid with subcutaneous ecchymosis and ptosis. Slit lamp examination on the left eye showed severe conjunctival edema, subconjunctival hemorrhage, corneal edema, Descemet’s folds, 3 + cells, 3 + Tyndall effect, iris surface hemorrhage with localized posterior synechiae, fibrin exudation on the lens surface. Indirect ophthalmoscopy revealed generally normal optic disc and retina, but details could not be seen clearly.
We gave the diagnosis of ASI due to ASC occlusion, followed by several treatments, including intravenous alprostadil for 10 days, subconjunctival injection (dexamethasone 2.5 mg/d) for 3 days and topical eye drops (prednisolone acetate 8 times/d). 12 days later, left anterior segment photography showed severe mixed conjunctival hyperemia, mild corneal edema, positive Tyndall effect, mid-dilated and irregular pupil (Fig. 1D). Fundus photography revealed no obvious abnormalities (Fig. 5A). Fundus fluorescein and indocyanine green angiography revealed normal retinal and choroidal artery perfusion in early-mid-late phase (Fig. 5B).
One month later, the best corrected visual acuity of the left eye improved to 20/40 from 20/200 at first visit. IOP was 20.2 mmHg OD and 13.5 mmHg OS. Slit lamp examination on the left eye showed transparent cornea, deep anterior chamber, negative Tyndall effect, mid-dilated and irregular pupil and iris atrophy of temporal quadrant. Two months later, the best corrected visual acuity of the left eye improved to 20/20. Left eye movement at any gaze and ptosis completely recovered.