Why the area under the eyes looks hollow, and how we approach the indian jureum (the Korean term for the tear-trough groove)

The groove under the eye is not a “wrinkle”. It is a boundary line where different tissues meet, and there is a structure holding that boundary in place.

Wong CH, Hsieh MKH, Mendelson B, Plastic and Reconstructive Surgery 2012;129(6):1392 — dissecting 48 cadaveric hemifaces, they consistently identified a true osteocutaneous ligament running straight from the maxilla to the skin. It passes between the two parts of the orbicularis oculi, beginning just below the anterior lacrimal crest and running roughly as far as the mid-pupillary line.

Since this ligament holds the skin towards the bone, when the orbital fat above is pushed forward and the cheek below descends, a groove is carved between them. A convex–concave–convex terracing arises, and when light comes in obliquely the concave place becomes a shadow.

More recently a micro-CT study (11 cadavers · 22 specimens) has come out holding that this ligament is “not a separate ligament but a complex continuous with the orbicularis retaining ligament (ORL)”, and a histological study refuting the very existence of a true ligament is also cited. The anatomical debate is not yet over. The reason we write even this down is that the diagram commonly used in explaining this area is not settled fact.

One commonly sees the explanation that “with age the bone around the eye dissolves and the orbit widens”. On looking up the original sources, the results were split by ethnicity.

That is, for Koreans the grounds for the explanation “because the bone dissolves” are weak. What then remains — changes in the position of the fat compartments and factors on the skin · muscle side come to take a relatively larger share.

This makes a difference in practice too. If shrinking bone were the cause, filling deeply would be the answer; but if the cause is that compartments have shifted, then where to fill and where to leave alone becomes far more important.

Darkness under the eyes is commonly divided into four kinds — pigmented · vascular · structural (shadow) · mixed. There is a study that applied this classification to actual patients and produced the proportions.

Huang YL, Chang SL, Ma L, Lee MC, Hu S, International Journal of Dermatology 2014;53(2):164 — 65 patients (mean age 38.9) were classified using diagnostic imaging, ultrasound and a 9-item assessment.

A pure single type came to only 22%. So a flat statement of the kind “you are the pigmented type, so whitening will do it” is statistically unlikely to be right. This is why in the consulting room we shine the light from several directions and look at you lying down and sitting up separately — the shadow component changes with posture and lighting, and the pigment component does not.

We should also be exact about the skin thickness. In a study that took full-thickness biopsies from 39 sites on the face of 10 cadaveric heads, the place where the dermis was thinnest was the medial upper eyelid at 0.759 mm (the thickest, the lower lateral nasal wall, was 1.97 mm). “Thinnest on the face” is as far as these data let us go, and the primary source for the commonly quoted “thinnest on the body, 0.5 mm” we did not find.

Among the procedures for this area, the one with the thickest body of data is hyaluronic acid. Let us look at two large collations.

Trinh LN, Grond SE, Gupta A, Facial Plastic Surgery 2022;38:228 — 23 studies · 2,048 patients (13 prospective · 9 retrospective), mean follow-up 13.7 months.

Puyana C & Montes JR, Journal of Clinical and Aesthetic Dermatology 2025;18(11):44 — retrospective, 155 patients (mean age 48), mean follow-up 785 days (about 26 months), 27G cannula in 82%, mean 0.45 mL per side.

An improvement of 1 grade in 105 patients (68%), an improvement of 2 grades in 22 (14%) — together, 82% were maintained in a state of raised grade. And the effect at 18 months was not statistically different from that at 6 months (p=0.57).

It is commonly said that “filler is gone in 6 months to 1 year”, but in this area it lasts longer than that. We take it to be because it is a site of little movement and slow metabolism.

Lasting long is not always only good news. The mean time to onset of delayed complications is reported at 16.8 months, and for discoloration there are cases at 52 months. It means this is not an area to fill generously on the assumption that “it will go away anyway”. It is as well to know along with this that among the 23 papers in the review, randomised controlled trials numbered 0.

What divides the result most in this area is not the product but which layer it goes into. It is confirmed in figures.

Stagliano S et al., Applied Sciences 2021;11(23):11489 — 9 studies · 830 patients. Above the periosteum in 454 (54.7%) versus a more superficial layer in 376 (45.4%).

Where it was placed above the periosteum the blue show-through (Tyndall) was almost 0, and in more superficial layers it exceeded 30%.

Why this should be so — it is a physical phenomenon. On passing very small particles, blue light scatters about 10 times more than red light. So gel that has gone in superficially under thin skin shows through with a bluish cast. It is why this area, where the skin is thinnest on the face, is particularly vulnerable.

Fortunately it can be reversed. It can be dissolved within 24 hours with 30 to 75 units of hyaluronidase, and this works even years later. If it has only just arisen, massage alone resolves a good deal of it. This is why we use the hyaluronic acid family first in this area — it is the only family that can be returned to how it was if you are not happy with it.

In one study, changing the protocol alone took the Tyndall from 17% to 0%. That said, there is no trial that randomly assigned the layer for comparison, so to put it exactly, “going deeper is favourable on the safety side” is as far as the evidence goes, and “it must be above the periosteum” is beyond it. There are also two studies reporting 98 to 100% improvement with injection into a superficial layer.

Another thing much used around the eyes is salmon-derived polynucleotide. We will write down what we verified, as it is.

Kim JH et al., Aesthetic Plastic Surgery 2022;46(4):1902 — 30 enrolled · 28 completed, 18 weeks, objective measurement with an Antera 3D camera.

• Grade of the wrinkles at the outer eye p<0.001
• Wrinkles · texture · pores · depressions · haemoglobin all p<0.05
• No significant change in melanin

That the haemoglobin index improved means there is room to expect something on the vascular dark circle side, and that melanin stayed as it was means there is no evidence for the pigmented type. Laid over the classification above (mixed type 78%), the natural conclusion is that PN alone does not improve every dark circle.

There is also one trial comparing it directly with non-cross-linked hyaluronic acid — randomised · double-blind · split-face, 27 patients, 3 sessions at 2-week intervals. On subjective assessment of improvement there was no difference between the two arms, while the rates of improvement in elasticity · hydration · roughness · pore volume were higher on the PN side.

There is something we should state honestly. An independent clinical trial paper on a “dedicated under-eye” polynucleotide product itself we did not find. The study above looked at the outer corner of the eye (crow's feet); it did not measure the volume of the groove under the eye. So we use this family not to fill the groove but to address the texture and the thinning of the skin. If the hollowing itself is the problem, a different approach is the right one.

Swelling that lasts

The commonest trouble in this area is not nodules but swelling.

Griepentrog GJ et al., American Journal of Cosmetic Surgery 2011;28(4):251 — on retrospective review of 51 cases from 4 practices, 12 patients (24%) had persistent periorbital oedema, lasting a mean of 5.4 months. 3 of the 12 improved with hyaluronidase.

In the review of 2,048 patients above, swelling was 35.6% of all adverse events, and in 5-year follow-up data malar edema was 11%. Here is why “a little bit more” is dangerous in this area.

The explanation “because lymphatic drainage is blocked” is widely used, but the primary evidence for that explanation we were not able to verify. More recently a case series has come out looking by ultrasound at the venous congestion hypothesis.

Blood vessels

Anatomically it is certainly a dangerous site. Jitaree B et al., Plastic and Reconstructive Surgery 2018;142(5):1153 — measurements on 30 cadaveric hemifaces.

• The nearest artery to the point at the lower border of the tear trough was the palpebral branch of the infraorbital artery, 2.79 ± 1.08 mm lateral · 2.88 ± 1.57 mm inferior
• At the point towards the medial canthus, the angular artery lay 4.00 ± 2.37 mm medially, and in these specimens it was confirmed as a branch of the ophthalmic artery — an anatomical route by which a retrograde embolus can reach the retina

That said, the distribution by site of the blindness cases actually reported is different. Among the 48 new cases in Beleznay K et al., Aesthetic Surgery Journal 2019;39(6):662, which gathered the world literature, the sites were nose 56.3% · glabella 27.1% · forehead 18.8% · nasolabial 14.6%, and the tear trough · under-eye does not appear as a separate item (1 upper eyelid, 2 cheek). In the two reviews above (2,048 patients · 830 patients) as well, vascular occlusion and blindness were 0 cases.

This must not be read as “under the eye is safe”. What has not been reported and what does not happen are different, and the anatomical route is real. The exact way to put it is “the anatomical risk is clear, but the blindness cases reported to date are concentrated in the nose · glabella · forehead · nasolabial fold”.

Clumping

Contour irregularity · nodules came to 7.57% (117/1,545) in the data on 2,048 patients, with wide variation between studies (33% in a study using a large-particle product). The proportion needing hyaluronidase was 5.18%. The choice of product and the amount change this number greatly.

The order we follow in the consulting room is simple.

1. We look with the lighting changed. What changes with posture and light is the shadow component; what does not change is the pigment component
2. We distinguish whether the fat has been pushed forward or the area has hollowed. If what has been pushed forward is the main problem, filling will not solve it and may make it look more swollen instead
3. We start with a small amount. The mean amount used is 0.45 to 0.47 mL per side, and in this area it lasts a long time
4. We use a reversible material first

The nasolabial fold is written up in The real reason nasolabial folds deepen, and how to improve them, and the difference between fillers and boosters in What is the difference between a filler and a booster.

Written and reviewed by Lee Chi-Hak, MD, medical director of Miso Clinic in Daegu, South Korea. Every study cited above is given together with its design, its size and the limitations the authors themselves recorded, and where we could not find data, we have said that we could not find any.