Collagenase Doses Tested in Rats Define Safety Window for Porcine Collagen Fillers

Cross-linked porcine collagen fillers withstand enzymatic degradation far better than their uncross-linked counterparts, and collagenase concentrations at or below 300 IU/mL can dissolve these fillers in living tissue without causing notable injury, according to a new open-access study published on 1 July 2026 in the journal Collagen and Leather. The work, led by Hang Wang, Rui Zhao and He Qiu together with colleagues, is presented by its authors as the first systematic in vitro and in vivo characterization of how two National Medical Products Administration-approved porcine type I collagen fillers behave under collagenase exposure, and of where the safety boundary for collagenase-based intervention lies.

The research team compared two formulations of porcine collagen filler, abbreviated PCF: a glutaraldehyde cross-linked version, CL-PCF, and an uncross-linked version, UCL-PCF. Both were 3.5% weight-per-volume suspensions of porcine type I collagen in phosphate-buffered physiological saline, supplied in sterile 1 mL syringes with 27-gauge needles. A commercial hyaluronic acid, or HA, filler served as the clinical benchmark throughout the study. The purified bacterial collagenase used in the experiments came from Clostridium histolyticum.

Material characterization revealed clear physicochemical differences rooted in crosslinking. Under the microscope, CL-PCF showed a compact, fibrous structure with uniform distribution of irregularly shaped fibrous gels, whereas UCL-PCF displayed noticeable clumping, higher viscosity and larger particle sizes. Rheological testing on an Anton Paar MCR 302 rheometer showed that all three fillers behaved as gel-like viscoelastic solids, with the elastic modulus exceeding the viscous modulus across all frequencies. Strain sweep tests indicated flow onset at 13.11% strain for UCL-PCF, 21.83% for CL-PCF and 18.63% for HA, highlighting CL-PCF’s superior resistance to deformation. Extrusion force measurements, however, revealed a trade-off: average push-out forces were significantly lower for UCL-PCF at 10.54 plus or minus 0.78 newtons and for HA at 9.98 plus or minus 1.43 newtons, compared with 14.48 plus or minus 1.23 newtons for CL-PCF, meaning the cross-linked material is more stable but harder to inject.

In vitro enzymolysis tests at 300 IU/mL collagenase showed that UCL-PCF could be mostly hydrolyzed within half an hour, while CL-PCF exhibited noticeable resistance for up to two hours, demonstrating that glutaraldehyde crosslinking markedly enhances collagen’s resistance to enzymatic breakdown. A 96-well microplate assay quantified CL-PCF degradation as time- and concentration-dependent: after 24 hours, the degradation rate reached 64.04% at 1200 IU/mL and 36.90% at 300 IU/mL. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis showed that bovine collagen and the porcine fillers displayed comparable sensitivity to collagenase, with the molecular weight of the enzyme measured at 63.54 plus or minus 0.87 kilodaltons. The authors caution that degradation efficiency may vary across collagen fillers depending on source, concentration, crosslinking technique, particle size, hardness and cohesiveness.

To probe vascular safety, the researchers used a rabbit ear embolization model. Twenty white New Zealand rabbits weighing 2.2 to 2.5 kilograms were randomly assigned to four groups of five, and 0.1 mL of filler was injected into the anterior branch of the posterior auricular artery. Both PCFs induced early partial intravascular obstruction with patterns distinctly different from HA. In the CL-PCF group, embolic material remained confined to the proximal artery without distal migration, while UCL-PCF, being more fluid and less degradation-resistant, showed partial washout and gradual breakdown in vivo. Peripheral capillary perfusion was preserved and no skin necrosis occurred in either PCF group. By contrast, the HA group developed marked ecchymosis by postoperative day 2, marginal ischemic necrosis by day 3 and progressive ulceration by day 7. By day 3, almost complete recanalization of the main central auricular artery occurred in all PCF-treated rabbits but in none of the HA-treated rabbits.

Histology supported these observations. Vessels in both PCF groups largely preserved their architecture, with an intact internal elastic lamina and continuous endothelium, no evident inflammatory infiltration, hemorrhage or perivascular edema, and only minimal focal intimal irregularity in occasional sections. HA-treated vessels showed intraluminal filler-associated occlusion with vascular wall injury and more obvious ischemic tissue damage. The authors propose mechanistic explanations: HA forms dense, homogeneous plugs that fully occlude small arteries and lacks intrinsic intravascular enzymatic clearance, whereas collagen-based fillers generate less cohesive emboli that may fragment under physiologic flow and are susceptible to endogenous proteolysis by matrix metalloproteinases, potentially shortening the duration of obstruction. They emphasize, however, that these results do not exclude the influence of injection speed, pressure fluctuation, dose volume and formulation differences, and that PCFs are not entirely devoid of vascular adverse event potential.

The in vivo enzymolysis assessment used 48 six-week-old male Sprague-Dawley rats weighing 180 to 200 grams, each receiving subcutaneous implantation of CL-PCF at two dorsal sites, one treated with sterile saline and the other with collagenase solution. Macroscopically, complete enzymolysis of 0.1 mL CL-PCF occurred within approximately 4 hours at 150 and 300 IU/mL and within 2 hours at 450 and 600 IU/mL, with the degradation process continuing for 24 to 48 hours. At 150 IU/mL, the main subcutaneous finding was vascular dilation without obvious tissue destruction. The 300 IU/mL group showed mild inflammatory infiltration and increased hemorrhage in surrounding tissue within one day. At 450 and 600 IU/mL, substantial inflammatory infiltration and varying degrees of skeletal muscle lysis appeared. Once concentrations reached 900 or 1200 IU/mL, severe epidermal damage occurred within 24 hours, accompanied by extensive matrix loss, hemorrhage and focal skeletal muscle lysis with inflammatory infiltration.

Quantitative analysis of skin collagen content based on Masson staining showed that, except for the 150 IU/mL group, collagen content in the 300, 450 and 600 IU/mL groups was significantly influenced by subcutaneous collagenase at certain time points, though no significant differences remained among the four groups seven days after injection. Dermal thickness in the mid-concentration groups returned to normal within one week, and macroscopic dorsal skin appearance showed no significant differences among the four lower-dose groups. The authors suggest that collagenase and its degradation products may contribute to anti-inflammatory processes, fibroblast and keratinocyte proliferation, and vascular regeneration, while noting that higher concentrations increase risks of tissue injury, discoloration and rare hypersensitivity.

From these data, the authors identify 300 IU/mL with an early 2-hour exposure window as a practical reference point for time-sensitive management of collagen filler-related vascular compromise in their models, consistent with the principle that earlier enzymatic reversal is associated with better perfusion recovery. They frame this as a preliminary benchmark rather than a standardized protocol, stressing a trade-off between dissolution speed and collateral tissue injury at excessive doses. In their conclusions, they report that collagenase at 150 IU/mL was well tolerated by surrounding tissues and that doses below 300 IU/mL were sufficient to correct collagen overfilling in the rat model, while the cross-linked filler’s architecture conferred greater tissue support and reduced migration.

The study carries substantial limitations that the authors detail explicitly. Only two PCF formulations and a single collagenase preparation were examined, limiting extrapolation to the broader landscape of collagen-based materials, since collagen source, purity, crosslink density and fibrillar microarchitecture can alter mechanical behavior, enzymatic susceptibility and host response. The in vitro assays cannot replicate the dynamic, enzyme-rich, immunologically active in vivo environment, so the findings should be read as comparative, clinically motivated evidence rather than definitive kinetic parameters. No standardized dosing or safety window exists for collagenase in intravascular contexts, and the rabbit ear model cannot fully recapitulate human facial arterial anatomy, vessel caliber, branching geometry or tissue compliance. Subcutaneous collagenase may also induce dose-dependent local inflammation and tissue injury requiring dedicated evaluation. All results reported are from rabbit and rat models and in vitro assays; the authors call for expanded multi-material, multi-enzyme and multi-model investigations before collagenase use in collagen filler complication management can be systematized in clinical practice.

Subject of Research: Enzymatic degradation behavior of porcine collagen fillers and collagenase safety thresholds in animal models

Article Title: Enzymolysis behavior and safe collagenase management of porcine collagen fillers

Article References: Wang, H., Zhao, R., Lo, C., Yu, D., Peng, H.-L. P., Wang, J., & Qiu, H. (2026). Enzymolysis behavior and safe collagenase management of porcine collagen fillers. Collagen and Leather, 8(1), Article 21. https://doi.org/10.1186/s42825-026-00244-x

Image Credits: AI Generated

DOI: 10.1186/s42825-026-00244-x

Keywords: porcine collagen filler, collagenase, crosslinking, glutaraldehyde, hyaluronic acid filler, vascular embolization, rabbit ear model, Sprague-Dawley rats, enzymatic degradation, soft-tissue augmentation, tissue safety, biomaterials