Severe wound complications following total knee arthroplasty (TKA), though uncommon, are of major importance. Clinical presentation ranges from wound problems and superficial infections to full-thickness skin necrosis (Galat, 2009). One of the possible causes of skin necrosis is the administration of low-molecular-weight heparin. This complication, occurring at a distance from the injection site, has been increasingly reported and more frequently observed with subcutaneous therapy, though it is an extremely rare event. Female gender, high BMI (> 25 kg m-2), and long duration of heparin therapy have been identified to be the risk factors; however, little is known about the true incidence (Schindewolf et al., 2009). These mechanisms are considered: immunologically mediated either via thrombosis resulting from heparin-induced immune aggregation of platelets (heparin-induced thrombocytopenia syndrome, HIT) or a formation of antigen–antibody complexes in cutaneous blood vessels (type III hypersensitivity syndrome) (Handschin et al., 2005). Moreover, to the best of our knowledge, there is only one case reported in the English literature about tissue necrosis occurring after knee arthroplasty (Karmegam and Raut, 2011). The authors describe a case of acute knee periprosthetic joint infection (PJI) complicated with severe skin necrosis in an anticoagulated patient that stopped dabigatran before the surgery and started enoxaparin on the post-operative period. PJI treatment required a two-stage approach with implant removal, medial gastrocnemius muscle flap, negative pressure dressing, hyperbaric oxygen chamber, and split-thickness skin graft. After laborious soft-tissue management, the patient ultimately underwent a successful second-stage procedure, being able to walk autonomously and presenting a good knee range of motion, with complete wound healing.

A 78-year-old woman, with a history of atrial fibrillation and hypertension, underwent left TKA for primary osteoarthritis (Fig. 1a). Chronic medication included dabigatran, olmesartan–hydrochlorothiazide, and amlodipine. There was no history of renal failure, diabetes mellitus, autoimmune disease, peripheral arterial or venous disease, or other relevant comorbidities (CHA2DS2-VASc = 4 points). Dabigatran was stopped 3 d before surgery. Throughout the procedure, a tourniquet was used (45 min with a pressure of 300 mm Hg), and a suction drain was in place for 24 h. During the post-operative period, therapeutic enoxaparin (60 mg BID) was started. The patient was discharged on day five with a clean wound and recommendation to switch back to oral dabigatran. On the 18th post-operative day, a significant wound skin necrosis was found with slight concurrent wound leakage (Fig. 1b), and despite C-reactive protein (CRP) levels of 15.8 mg L-1, there was a high suspicion of an underlying acute periprosthetic joint infection. The patient was switched back to enoxaparin and was scheduled for surgery within 2 d (Fig. 1c). After surgical debridement with implant retention (DAIR), broad-spectrum IV antibiotics were started (vancomycin and piperacillin–tazobactam). Despite these efforts, skin necrosis progressed on the distal part of the wound (Fig. 2a). Microbiological samples confirmed infection with Enterococcus faecalis and Staphylococcus haemolyticus in multiple samples. On the fifth day after DAIR, skin necrosis at the distal part of the wound progressed (Fig. 2a), and a decision was made to remove the prosthesis and implant an antibiotic-loaded handmade mobile spacer (with 4 g vancomycin and 2 g meropenem per bag of bone cement) and perform a medial gastrocnemius muscle flap for wound coverage (Fig. 2b–c). Microbiological samples, including implant sonication, isolated the same microorganisms, but given the unfavourable clinical course, broad-spectrum antibiotics were maintained for 6 weeks.

During the next weeks, an extensive superficial skin necrosis developed around the knee, extending to the lower leg around the incision made to harvest the muscle flap (Fig. 3). An arteriography was performed, and occlusive arterial disease was definitively ruled out. Skin biopsy revealed microvascular thrombotic phenomena and ischemia without vasculitis. Complete blood count (CBC) revealed persistently normal platelet values and slight post-operative anaemia. Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels were elevated but ran a favourable downward trend. All coagulation parameters were normal, and autoimmune disease assessment was negative. A haematology consult was also requested, thinking of low-molecular-weight heparin (LMWH)-induced skin necrosis, but given the lack of thrombocytopenia and atypical clinical presentation, a decision was made not to pursue more specific testing regardless of our suspicion.

After 3 weeks, surgical debridement of the necrotic tissues was undertaken. With the lack of an alternative diagnosis, a decision was made to discontinue enoxaparin and reinstate dabigatran. A negative-pressure dressing was applied, and complementary treatment on a hyperbaric oxygen chamber was also initiated. Necrosis stopped progressing (Fig. 4a–d), and eventually a split-thickness skin graft of the thigh for wound coverage was performed (Fig. 4e).

During the outpatient follow-up, soft tissues showed slow but progressive improvements and eventually healed completely. A few weeks after hospital discharge, an ulcer on the lateral aspect of the leg (unrelated to previous incisions) developed and progressed (Fig. 5a–d). A biopsy was again performed that revealed unspecific findings, suggesting an infected venous leg ulcer, and a 2-week course of oral amoxicillin–clavulanate was instituted with excellent clinical response (Fig. 5e–h). A new immunological assessment showed positive ANA with a low value of anti-dsDNA, without complement consumption and positive IgG anti-beta-2-glycoprotein I (anti-cardiolipin and lupus anticoagulant remained negative), but after 12 weeks they were back to normal.

Second-stage surgery was consecutively postponed, but when satisfactory soft tissues were present (Fig. 6a) and after an adequate discussion with the patient, a decision to go ahead with the second stage was taken 1 year after the original surgery. Dabigatran was discontinued for 3 d before, without enoxaparin bridging, and immediately resumed after surgery. Intraoperatively, multiple deep tissue samples were collected and showed no bacterial growth. A period of 2 years after the second stage, the patient is infection-free, with a painless, well-functioning knee arthroplasty, and no systemic disorder was diagnosed (Fig. 6).

Extensive skin necrosis can originate in a wide range of etiologies, detailed in the following subsections.