In Oscar Wilde's 1890 novel, Dorian Gray keeps a painted portrait locked in an attic room while his own face stays unchanged by decades of vanity and cruelty; the canvas absorbs every mark instead. When Dorian finally can't bear the evidence anymore, he stabs the portrait through the heart, meaning to destroy the one thing that has recorded what he actually did. The undoing is instant and total: servants find a beautiful young portrait restored to its first state, and on the floor beneath it an old, disfigured stranger wearing Dorian's rings. Wilde's fantasy is that consequence can be sealed into an object and then erased in a single clean stroke, no matter how long it had been accumulating.

That fantasy is, more or less, what a lot of patients are quietly hoping for when they ask about dissolving a hyaluronic acid filler. "If I don't like it, you can just dissolve it" gets said in consultations the way Dorian might have described his attic arrangement — a private undo button, instantaneous, complete, waiting behind a locked door. Hyaluronidase is real and it does dissolve HA filler. It is not, however, a knife through canvas. What actually happens once the enzyme goes in depends heavily on what kind of "portrait" is being erased, and the honest research on that question is smaller and more uneven than the confidence of the sales pitch suggests.

What the Enzyme Actually Does

Hyaluronidase works by hydrolyzing the glycosidic bond between N-acetylglucosamine and glucuronic acid — the two sugars that repeat to form the hyaluronic acid backbone (Paap & Silkiss, 2020). That's a precise, narrow chemical action, and it runs into a problem HA filler manufacturers built on purpose: the same cross-linking that lets a filler outlast the body's own hyaluronic acid by months is the same structural property that slows an enzyme trying to cut it apart. Paap and Silkiss's 2020 comparative review found that filler resistance to hyaluronidase tracks predictably with three variables — HA concentration, cross-linking density, and whether the gel is monophasic or biphasic (more cross-linked, monophasic gels expose less surface area to the enzyme). In their comparison, the Juvéderm family, built as densely cross-linked monophasic gels, came out consistently more resistant to breakdown than the more loosely cross-linked, biphasic Restylane family, with Belotero's results falling inconsistently in between depending on the study. In other words: the portrait isn't painted in one universal medium. Some are oil, some are watercolor, and the knife meets very different resistance depending on which one is on the wall.

How Long "Instant" Actually Takes

Scrima et al. (2022) put this to a direct test, combining an in vivo mouse model with in vitro degradation assays tracked by high-frequency ultrasound across four HA-based products of increasing structural complexity: a simple linear HA filler, two HA "hybrid complex" formulations, and a cross-linked gel. Before hyaluronidase even entered the picture, the products' natural persistence already varied enormously — the linear HA filler was undetectable by four weeks, the lower-concentration hybrid complex lasted about ten weeks, the higher-concentration hybrid complex lasted twenty-nine weeks, and the cross-linked gel lasted thirty-three weeks. Then they added hyaluronidase at a fixed concentration and watched what happened: the hybrid complex reached near-complete, homogeneous depolymerization within three hours, while the cross-linked gel needed a full twenty-four hours to reach a comparable state, and critically, only about 13% of the cross-linked gel was even in a soluble, enzyme-accessible fraction at the start, compared to 100% for the hybrid complex. The filler that was engineered to last the longest in the skin is, unsurprisingly, also the one that resists being undone the longest once a patient decides she wants it gone. Dorian's portrait absorbed decades in a single object and let go of them in a single stab; a cross-linked dermal filler took months to earn its resistance and needs the better part of a day to give it back, if it gives it back evenly at all.

The Evidence Gap Nobody Advertises

Here is the part of the story that belongs in Dorian's locked room. Borzabadi-Farahani et al. (2022) conducted a scoping review searching PubMed, Medline, Embase, and Cochrane through May 2022, screening 395 identified studies for anything resembling controlled evidence on hyaluronidase reversal. They found five randomized controlled trials — a combined total of 53 subjects — and every one of them studied uncomplicated nodule removal in non-facial sites: the forearm, the upper arm, the back. Reported single-injection dosing across the literature ranged from 1.25 to 37.5 units per 0.1 mL of filler, or, in fractionated protocols, 0.375 to 2.25 units per injection repeated weekly over three weeks, with no major adverse reactions recorded in the trials reviewed. What the reviewers explicitly could not find was any randomized, controlled evidence — none — for reversing HA filler in facial skin, which is where the overwhelming majority of real-world hyaluronidase injections actually happen. The dose a patient's face receives when she wants her lips softened is, more often than clinicians like to admit, extrapolated from data collected on someone's forearm.

When Undoing Is Not Optional

The stakes change entirely when hyaluronidase is being used not for regret but for an emergency — a filler accidentally injected into or compressing a blood vessel, cutting off circulation to skin or, in the worst cases, to the eye. Alhusain et al. (2026) pooled 21 studies covering 231 patients (mean age 30.1, 90.9% female) treated for exactly this scenario and found that among the complications reported, skin necrosis occurred in 60.6% of cases and visual impairment in 52.8%. Hyaluronidase resolved the necrosis in 78.4% of those cases — a reassuring number — but resolved visual impairment in only 45.2%, even with treatment. The variable that mattered most was time: intervention within six hours of the vascular event significantly improved outcomes, and higher hyaluronidase doses correlated with better necrosis resolution specifically. This is the real version of the Dorian Gray fantasy, and it inverts the novel completely. There is no single stroke that instantly restores what was lost. There is a narrowing window, measured in hours, in which enough enzyme delivered quickly enough can save tissue — and even then, for the eye, less than half the time.

Wilde wrote a story about consequence hidden away and erased all at once, cleanly, on demand. The actual chemistry of undoing a dermal filler runs the other direction on every count that matters: what dissolves easily was designed to dissolve easily, what was built to last resists being undone in proportion to how well it was built, the evidence for facial reversal is thinner than the confidence with which it's offered, and when the undoing is truly urgent, it is a race measured in hours rather than a single clean cut — one that succeeds more often for skin than it does, even now, for sight. There is no locked attic where a filler's history sits waiting to be stabbed away in an instant. There is only an enzyme, a molecule it was or wasn't built to resist, and a clock.

References

Alhusain, A. M., Alsaif, M., Alkhathami, A. M., Alghamdi, A. M., Alkhunein, J., Alqirnas, M., Alsubhi, R. O., Alsalamah, S., Alanazi, M. M., Fathi, A., Alghamdi, A., Alowid, F., & Aleissa, S. (2026). Hyaluronidase protocol in management of hyaluronic acid filler-related vascular complications: A systematic review and meta-analysis. Aesthetic Plastic Surgery, 50(6), 2300–2317. https://doi.org/10.1007/s00266-025-05431-5

Borzabadi-Farahani, A., Mosahebi, A., & Zargaran, D. (2022). A scoping review of hyaluronidase use in managing the complications of aesthetic interventions. Aesthetic Plastic Surgery, 46(6), 1193–1209. https://doi.org/10.1007/s00266-022-03207-9

Paap, M. K., & Silkiss, R. Z. (2020). The interaction between hyaluronidase and hyaluronic acid gel fillers: A review of the literature and comparative analysis. Plastic and Aesthetic Research, 7, Article 36. https://doi.org/10.20517/2347-9264.2020.121

Scrima, M., Merola, F., Vito, N., Pacchioni, D., Vecchi, G., Melito, C., Iorio, A., Giori, A. M., & Ferravante, A. (2022). Elucidations on the performance and reversibility of treatment with hyaluronic acid-based dermal fillers: In vivo and in vitro approaches. Clinical, Cosmetic and Investigational Dermatology, 15, 2629–2640. https://doi.org/10.2147/CCID.S383354