Autogenous bone grafting limitations have motivated the development of Tissue-Engineered (TE) biomaterials that offer an alternative as bone void fillers. However, the lack of a blood supply within implanted constructs may result in avascular necrosis and construct failure. 1. The aim of this project was to investigate the potential of novel TE constructs to promote vascularisation and bone defect repair using two distinct approaches. In Study 1, we investigated the potential of a mesenchymal stem cell (MSC) and endothelial cell (EC) co-culture to stimulate pre-vascularisation of biomaterials prior to in vivo implantation. 2. In Study 2, we investigated the potential of TE hypertrophic cartilage to promote the release of angiogenic factors such as VEGF, vascular invasion and subsequent endochondral bone formation in an in vivo model. Collagen-only (Coll), collagen-glycosaminoglycan (CG) and collagen-hydroxyapatite (CHA) scaffolds were fabricated by freeze-drying. 3. , seeded with cells and implanted into critical-sized calvarial and femoral defects in immunocompetent rats. In Study 1, Coll and CG scaffolds were initially seeded with ECs, allowed to form capillary-like networks before the delayed addition of MSCs and continued culture prior to calvarial implantation. In Study 2, CG and CHA scaffolds were seeded with MSCs and cultured under chondrogenic and subsequent hypertrophic conditions to form a cartilage pre-cursor prior to calvarial and femoral implantation in vivo. MicroCT and histomorphometry quantification demonstrated the ability of both systems to support increased bone formation compared to controls. Moreover, the greatest levels of bone formation were observed in the CG groups, notably in those containing cartilage tissue (Study 2). Assessment of the immune response suggests the addition of MSCs promotes the polarisation of macrophages away from inflammation (M1) towards a pro-remodelling phenotype (M2). We have developed distinct collagen-based systems that promote vascularisation and ultimately enhance bone formation, confirming their potential as advanced strategies for bone repair applications

1. The rates of vascularisation in 119 autogenous, homogenous and heterogenous bone grafts, placed in the femoral medullary cavity and under the renal capsule of rabbits, were studied. 2. Substantial differences have been found in the speed of vascular penetration and arrangement among autografts, homografts and heterografts : penetration of the heterogenous implant was six or more times slower. Moreover, large areas of the homografts and heterografts were often totally excluded from the circulation for as long as the research was continued (up to three months). Revascularisation of the cortical bone was slower and less profuse than in cancellous bone, keeping always the same respective proportion between the three types of bone we have described. The results on the kidney were much less constant, and I attribute this to the vascular peculiarities of the bed. 3. Vascular patterns peculiar to the time of implantation and type of graft are described. 4. Suggestive, even if not totally convincing, evidence was found of recanalisation of old vessels inside the graft by advancing vessels from the bed. 5. There is striking correlation between the rate of vascular penetration of the bone implants and their ultimate "take" or incorporation in the bed

Purpose: To evaluate the vascularisation of the femoral head in children with slipped capital femoral epiphysis (SCFE) before and after surgery with use of contrast-enhanced MRI. Methods and Materials: 20 consecutive children, 13 boys and 7 girls, aged 9–15 years, with slipped capital femoral epiphysis, were included into the study. The classification of SCFE was performed traditionally due to the patient’s history, physical examination and findings of the radiographs. There were no pre-slips, 9 children had acute, 5 children had acute-on-chronic and 6 children had chronic SCFE. The MRI-examinations were performed in a 1.5 Tesla MR-scanner with use of the body coil and all postoperative MR-examinations were carried out within 4 weeks after surgery. The examination protocol included a coronal fat-suppressed STIR-sequence, a coronal contrast-enhanced T1-weighted spin-echo sequence and a sagittal 3D-gradient-echo (FFE) sequence. Morphology, signal intensities and contrast-enhancement of the femoral head were assessed retrospectively by two experienced radiologists in consensus. Results: Morphologic distortion of the physis, bone marrow edema in the metaphysis and epiphysis and joint effusion were the preoperative MRI-findings of slipped capital femoral epiphysis in each child. In 17 children, who underwent in situ-fixation with a single screw, and in one child, who underwent open reduction of the epiphysis, the vascularisation of the femoral head before and after surgery was normal. An avascular zone in the posterior-lateral aspect of the epiphysis was visible preoperatively in one child, which completely revascularized after open reduction and internal fixation of the epiphysis with two screws. One child with severe SCFE developed avascular necrosis of the femoral head after open reduction and corrective osteotomy through the physis. Conclusion: MRI allows for accurate evaluation of the femoral head vascularisation before and after surgery in children with slipped capital femoral epiphysis

1. Serial radiographs of fifty-two normal children's feet, taken at six-monthly intervals between two and five years, have been reviewed. 2. Twenty-one naviculars have been injected post-mortem and the vascularisation of the growing bone investigated. 3. The records of sixty-two children with a diagnosis of Köhler's disease have been studied. 4. It is submitted that abnormal ossification results from compression of the bony nucleus at a critical phase during growth of a navicular bone whose appearance is delayed. 5. Symptoms in Köhler's disease are related to further compression which produces vascular changes in the bony nucleus. Consequent ischaemia is followed by hyperaemia which produces local pain, tenderness and swelling. 6. Two radiographic types of Köhler's disease are described and attributed to variations in the basic vascular pattern of the affected bone. 7. The usual complete recovery of the navicular is ascribed in part to the basic arrangement of numerous radially penetrating vessels

The femoral head receives its blood supply primarily from the medial femoral circumflex artery, with its deep branch being the most important.

In a previous study, we performed classical anatomical dissections of 16 hips. We have extended our investigation with a radiological study, in which we aimed to visualise the arteries supplying the femoral head in healthy individuals. We analysed 55 CT angiographic images of the hip.

Using 64-row CT angiography, we identified three main arteries supplying the femoral head: the deep branch of the medial femoral circumflex artery and the posterior inferior nutrient artery originating from the medial femoral circumflex artery, and the piriformis branch of the inferior gluteal artery. CT angiography is a good method for visualisation of the arteries supplying the femoral head. The current radiological studies will provide information for further investigation of vascularity after traumatic dislocation of the hip, using CT angiography.

Summary Statement

Prolonged presence of VEGF (released from gelatin microspheres) led to a significant increase in scaffold vascularization when applied in vivo. Bioprinted scaffolds with regional VEGF presence retained their architecture and regional vessel formation occurred.

In this study the direct relationship between the type of bone implant used, the vascular reaction caused to the host and the revascularisation of the implant has been studied. It was found that the best graft was that which was the most rapidly and permanently vascularised. Not only was the biological affinity between the graft and the bed important, but the structural facilities offered by the implant for the "penetration" by the host vessels were also of paramount importance. Thus small, fresh, cancellous bone grafts offered the best chance of rapid incorporation provided they were not crushed to the point of making vascular progress difficult. The findings from this investigation so strongly suggest that the rapid revascularisation of the bone grafts was because of an end-to-end anastomosis of the vessels of the host with those in the implant that it seems justified to consider that the best bone graft is that which is richest in vessels. Apart from a recent short paper by Graf (1960), we have not found this assertion before. It is this which seems to make the fresh, autogenous, cancellous implant so superior to all others.

We believe that any new material for bone grafts should be tested by the technique described here. The material which one day may replace fresh, autogenous, cancellous implants will have to show the same readiness to vascular penetration, vascular osteogenesis and vascular permanency that at present is exhibited only by the cancellous autograft.

We performed a series of 16 anatomical dissections on Caucasian cadaver material to determine the surgical anatomy of the medial femoral circumflex artery (MFCA) and its anastomoses. These confirmed that the femoral head receives its blood supply primarily from the MFCA via a group of posterior superior nutrient arteries and the posterior inferior nutrient artery. In terms of anastomoses that may also contribute to the blood supply, the anastomosis with the inferior gluteal artery, via the piriformis branch, is the most important. These dissections provide a base of knowledge for further radiological studies on the vascularity of the normal femoral head and its vascularity after dislocation of the hip.

Critical-sized bone defects remain challenging in the clinical setting. Autologous bone grafting remains preferred by clinicians. However, the use of autologous tissue is associated with donor-site morbidity and limited accessibility to the graft tissue. Advances in the development of synthetic bone substitutes focus on improving their osteoinductive properties. Whereas osteoinductivity has been demonstrated with ceramics, it is still a challenge in case of polymeric composites. One of the approaches to improve the regenerative properties of biomaterials, without changing their synthetic character, is the addition of inorganic ions with known osteogenic and angiogenic properties. We have previously reported that the use of a bioactive composite with high ceramic content composed of poly(ethyleneoxide terephthalate)/poly(butylene terephthalate) (1000PEOT70PBT30, PolyActive, PA) and 50% beta-tricalcium phosphate (β-TCP) with the addition of zinc in a form of a coating of the TCP particles can enhance the osteogenic differentiation of human mesenchymal stromal cells (hMSCs) (3). To further support the regenerative properties of these scaffolds, inorganic ions with known angiogenic properties, copper or cobalt, were added to the coating solution. β-TCP particles were immersed in a zinc and copper or zinc and cobalt solution with a concentration of 15 or 45 mM. 3D porous scaffolds composed of 1000PEOT70PBT30 and pure or coated β-TCP were additively manufactured by 3D fibre deposition. The osteogenic and angiogenic properties of the fabricated scaffolds were tested in vitro through culture with hMSCs and human umbilical vein endothelial cells, respectively. The materials were further evaluated through ectopic implantation in an in vivo mini-pig model. The early expression of relevant osteogenic gene markers (collagen-1, osteocalcin) of hMSCs was upregulated in the presence of lower concentration of inorganic ions. Further analysis will focus on the evaluation of ectopic bone formation and vascularisation of these scaffolds after implantation in a mini-pig ectopic intramuscular model

Introduction. Autologous fat grafting has favourable potential as a regenerative strategy and is the current gold-standard to repair large contour defects, as needed in breast reconstruction after mastectomy and traumatic soft tissue reconstruction. Clinically, there is a limit on the volume of lipoaspirate which can be utilised to repair a soft-tissue defect. Surgical complications are the result of poor structural fidelity of lipoaspirate and graft resorption as a filling material and are hindered further by poor graft vascularisation. This study aims to develop injectable lipoaspirate-derived adipose tissue grafts with enhanced biologically and clinically-admissible structural and functional properties adopting light photocrosslinking of unmodified lipoaspirate. Methods. Patient-derived lipoaspirate was harvested and crosslinked using novel photoinitiator and exposure to visible light (wavelength 450nm) in surgery, establishing bonds between extracellular matrix (ECM) proteins within the material. The degree of crosslinking was tuned (photoinitiator concentration, light exposure, light intensity) and covalent bond formation measured using mass spectrometry. To predict patient response, SWATH-MS was used to identify differences in patient ECM and crosslinked grafts were implanted in vivo using a subcutaneous mouse model. Functional vessel formation and resorption were quantified using micro-CT and tissue-remodelling was assessed via histology. Results. There was an increase in the relative abundance of covalent bonds present with increasing degree of crosslinking. When injected, crosslinked lipoaspirate had better shape fidelity compared with native lipoaspirate – demonstrated by a smaller fibre diameter. Crosslinked lipoaspirate remained viable over long term culture and resulted in more predictable resorption profiles when implanted in vivo. Conclusions. The crosslinking approach described here is tunable and functional across different patient samples. Improving the structural properties of lipoaspirate through minimal manipulation has clinical utility for the delivery of grafts with higher shape fidelity and therefore increased graft survival when implanted

Neoangiogenesis drives the replacement of mineralised cartilage by trabecular bone during bone growth regulated by molecules like e.g. VEGF, OPG and RANKL. The Heparan sulfate proteoglycan Syndecan-1 (Sdc1) plays a role in the interaction of osteoclasts and osteoblasts and the development of blood vessels. We expected Sdc1 to have an influence on bone structure and vessel development. Therefore, bone structure and angiogenesis at the growth plate in mice was compared and the influence of Syndecan-1 deficiency was characterised. Animals: Femura of male and female C57BL/6 WT (5♀, 6♂) and Sdc1-/- (9♀, 5♂) mice were used for native bone analysis at 4 month age. Histology: Bone structure was analysed using microCT scans with a resolution of 9µm. Vascularisation was visualised using an anti-Endomucin antibody in 80µm thick cryosections. In vitro angiogenesis: Bone marrow isolates were used to generate endothelial progenitor cells by sequential cultivation on fibronectin. Microvessel development was analysed 4h after plating on matrigel. Bone structure in male Sdc1 deficient mice was significantly reduced compare to male WT, whereas female mice of both genotypes did not differ. Sdc1 deficient mice at the age of 4 month showed a high decrease in the number of vessel bulbs at the chondro-osseous border (growth plate) compared to WT mice. However, no sex related differences were shown. Quantification of microvessel outgrowth of endothelial cells revealed a decreased amount of sprouting, but increased length of microvessels of Sdc1-/- cells compared to WT. Syndecan-1 has a significant impact on neoangiogenesis at the chondro-osseous border of the native bone, but the impact of Syndecan-1 deficiency on the loss of bone structure was significantly higher in male mice. This emphasises the importance to further characterise the function of Syndecan-1 regulated processes during enchondral ossification in a sex dependent manner

Introduction: The lumbosacral medulla is vascularized by the Adamkiewicz arteria which irrigates the anterior spinal arteria. Occlusion or section of the Adamkiewicz arteria may induce an ischemia of the medulla during anterior or transforaminal spine surgery. An angiography allows to determine the exact topography of this artery. The purpose of this study was to describe its preoperative topography and to analyze the impact of angiography on the surgical strategy. Methods: In this retrospective study, 100 preoperative medullar angiographies, performed by a vascular radiologist between january 1998 and august 2007, were reviewed. Surgical indications were: 50 vertebrectomies in tumors, 20 anterior fusions in dorsolumbar fractures, 10 anterior fusions in malunions, 10 anterior releases in scoliosis, 3 transpedicular osteotomies, 7 disc hernias (T7-L4). The level and the side of foraminal entrance of the Adamkiewicz arteria and collateral arterias irrigating the anterior spinal arteria were analyzed. We looked for the occurence of postoperative ischemic signs of the medulla. Modifications of surgical planning because of Adamkiewicz’ arteria topography were noted. The possibilities of preoperative tumor embolisation were analyzed. Results: The Adamkiewicz arteria was always localized between T8 and L3. It was present at the foraminal levels L1/L2 or L2/L3 in 48% of the cases. The left side was concerned in 65% of the cases. A modification of the surgical strategy was noted in 16% of the cases: 12 side changements of operative approach, 4 contra-indications for anterior surgery. An ischemic syndrome of the anterior lumbosacral medulla were not found. In the group of tumors, the preoperative angiography allowed to perform a selective embolisation of tumor vessels in 80% of the cases. In all other cases, the tumor vascularisation was common with the vascularisation of the medulla which could have made the embolisation dangerous. Conclusion: Although the occurence of a lumbosacral medullar ischemia secondary to an Adamkiewicz arteria lesion is rarely reported in the literature, the preoperative angiography reduces this potential risk. The exact knowledge of the anterior medullar vascularisation allows to better plan the surgical strategy and to adapt the side of operative approach. Furthermore, the angiography enables to perform a selective embolisation of tumors safely

Purpose: We report an retrospective analysis of 33 patients with neurological para-osteo-arthroplathy of the hip who underwent surgery between 1985 and 1999. Material and methods: Forty-three hips were operated in 33 patients aged 14 to 50 years at the time of the accident. Twenty-two patients had head trauma, two had spinal cord injury and three both. The causal mechanism was: trauma 27 patients, rupture of an inta-cranial aneurysm in five patients, widespread burns one patient. Localisations were inferome-dial 14 patients, anterior 10 patients, posterior two patients, circumferential five patients. Surgical care included complete resection in 30 cases, resection of the head and neck in six, and implantation of a total hip arthroplasty in seven. There were several perioperative accidents: two vessel injuries, two persistent bleedings, one haematoma, eight superficial infections, six recurrences, one ankylosis and one death. Functional outcome was assessed on the basis of gain in amplitude of hip flexion. Results: Analysis was possible for 37 of the 43 hips. Outcome was good in 18 (flexion gain greater than 90°), fair in nine (flexion gain from 60 to 90°), poor in ten (flexion gain less than 60°). Complete resection gave better results (61.5%). Total hip arthroplasty gave mediocre results. The best results were obtained with resection for the inferomedial and anterior localisations. Discussion: The decision for surgery should be discussed in light of the objectives to be achieved. We present our surgical strategy as a function of the localisation. We advocate systematic verification of the vascularisation and prefer the obturator approach for inferomedial localisations. The risk of recurrence depends on the delay from the accident to surgery. Conclusion: Based on this retrospective series, we prefer surgical resection after the first year. We emphasise the importance of peroperative verification of the vascularisation

Objectives. An experimental piglet model induces avascular necrosis (AVN) and deformation of the femoral head but its secondary effects on the developing acetabulum have not been studied. The aim of this study was to assess the development of secondary acetabular deformation following femoral head ischemia. Methods. Intracapsular circumferential ligation at the base of the femoral neck and sectioning of the ligamentum teres were performed in three week old piglets. MRI was then used for qualitative and quantitative studies of the acetabula in operated and non-operated hips in eight piglets from 48 hours to eight weeks post-surgery. Specimen photographs and histological sections of the acetabula were done at the end of the study. . Results. The operated-side acetabula were wider, shallower and misshapen, with flattened labral edges. At eight weeks, increased acetabular cartilage thickness characterised the operated sides compared with non-operated sides (p < 0.001, ANOVA). The mean acetabular width on the operated side was increased (p = 0.015) while acetabular depth was decreased anteriorly (p = 0.007) and posteriorly (p = 0.44). The cartilage was thicker, with delayed acetabular bone formation, and showed increased vascularisation with fibrosis laterally and focal degenerative changes involving chondrocyte hypocellularity, chondrocyte cloning, peripheral pannus formation and surface fibrillation. . Conclusions. We demonstrate that femoral head AVN in the young growing piglet also induced, and was coupled with, secondary malformation in acetabular shape affecting both articular and adjacent pelvic cartilage structure, and acetabular bone. The femoral head model inducing AVN can also be applied to studies of acetabular maldevelopment, which is less well understood in terms of developing hip malformation. Cite this article: Bone Joint Res 2014;3:130–8

Purpose: Reconstruction after resection of malignant bone tumours remains a major challenge. Free vascularised fibular grafts may be a useful alternative in this indication. Material: Thirty children (nine girls, twenty-one boys) were treated between 1993 and 2000. Mean age was eleven years. Tumour histology was: osteogenic osteo-sarcoma (n=20), Ewing tumour (n=5), justacortical osteosarcoma (n=3), synovialosarcoma (n=1), and chondrosarcoma (n=1). Tumours were located in the femur (n=17), the tibia (n=6), the humerus (n=5), the radius (n=1), and the distal fibular (n=1). The length of resection varied from 100 mm to 260 mm (mean 160 mm). Internal fixation was used in 27 cases and external fixation in three. The adjacent epiphysis was preserved in 22 cases and initial arthrodesis was performed in eight. Method: Patients were followed clinically and radiographically. A bone scintigram was obtained in all patients at least once during the postoperative period. Radiological assessment was based on the hypertrophy index of the graft using the method described by DeBoer and Wood. Functional outcome was assessed using the Enneking criteria. Results: Mean follow-up was 51 months (range 2 – 9 years). Early amputation was required in two patients due to local ocological complications. One patient died at eight months follow-up due to lung metastasis. Among the remaining 27 patients, primary bone healing was achieved in 22. The five other patients exhibited clear signs of non-vascularisation. Successful healing was achieved in four of these patients after a complementary autologous graft. All cases of stress fracture healed after simple immobilisation. The twenty-two patients who achieved primary bone healing developed a hypertrophic graft (mean 61%, range 22 – 190%). Graft hypertrophy was not observed in the five cases requiring a secondary graft after the scintigram demonstrated lack of vascularisation. Hypertrophy of the vascularised fibular graft was more marked for lower limb reconstructions than for upper limb reconstructions. Functional outcome was satisfactory in all cases. On the 30-point Enneking scale as modified, our patients achieved a mean 26 points (range 19 – 30). Discussion: Outcome was directly related to patency of the vascular anastomoses. Bone scintigraphy, performed one month after reconstruction surgery, was an important element for assessing prognosis. In case of unsuccessful vascularisation, a complementary cortico-cancellous graft should be used. Early weight-bearing is advisable using adequate protective devices. Dynamic osteosynthesis systems should be helpful in improving graft hypertrophy

Previous clinical studies have shown the efficacy of a foreign body-induced membrane combined with bone autograft for the reconstruction of traumatologic or pathologic large bone defects or, bone non union. This membrane, rich in mesenchymal stromal cells (MSC), avoids bone autograft resorption and promotes consolidation by revascularisation of the bone and secretion of growth factors. Reconstruction requires two different surgical stages: firstly, insertion of a cement spacer in the defect, and secondly, removal of the spacer, preservation of the foreign body-induced membrane and filling of the cavity by bone autograft. The optimal time to perform the second surgical stage remains unclear. So, we aimed to correlate bone healing and, phenotype and function of cells isolated from the induced membrane, in patients whose second surgery was performed on average after 6 months (i.e. beyond the recommended time of one month). Cell phenotype was determined by flow cytometry and cell function by: alkaline Phosphatase enzyme activity, secretion of calcium and von Kossa staining. Second, using histological and immunohistochemistry studies, we aimed to determine the nature and function of induced membrane over time. Seven patients were included with their consent. Results showed Treated patients achieved in all cases bone union (except for one patient) and in in vitro and histology and immunohistochemistry gave some indications which need to be completed in the future. First, patient age seemed to be an indicator of bone union speed and recurrent infection, appeared to influence in vitro MSC osteogenic potential and induced membrane structure. Second, we reported, in bone repair situation, the commitment over time in osteogenic lineage of a surprising multipotent tissue (induced membrane) able of vascularisation/ osteogenesis/ chondrogenesis at a precocious time. Finally, best time to perform the second stage (one month) could be easily exceeded since bone union occurred even at very late times

Background. Tissue engineering strategies to heal critical-size bone defects through direct bone formation are limited by incomplete integration of grafts with host bone and incomplete vascularisation. An alternative strategy is the use of cartilage grafts that undergo endochondral ossification. Endochondral cartilages stimulate angiogenesis and are remodeled into bone, but are naturally found in only small quantities. We sought to develop engineered endochondral cartilage grafts using human osteoarthritic (OA) articular chondrocytes. Methods. Study approval was obtained from our human and animal ethics review committees. Human OA cartilage was obtained from discarded tissues from total knee replacements. Scaffold-free engineered grafts were generated by pelleting primary or passaged chondrocytes, followed by culture with transforming growth factor-β1 (TGF-β1) and bone morphogenetic protein 4. Samples were transplanted into immunocompromised mice either subcutaneously or into critical-size tibial defects. Grafts derived from passaged chondrocytes from either of two patients (64 year old and 68 year old men) where implanted into tibial defects in five mice. Bone formation was assessed with histology after four weeks of implantation. Results. Engineered cartilage grafts generated from passaged OA chondrocytes underwent endochondral ossification after implantation either subcutaneously or in bone. The grafts bridged tibial defects, integrating with bone proximally and distally in all cases. Portions of the graft were remodeled into woven bone, which spanned the defects in two animals. Unmodified OA cartilage and engineered grafts formed from primary chondrocytes did not undergo endochondral ossification in vivo. Conclusions. Human OA chondrocytes adopt an endochondral phenotype after passaging and TGF-β superfamily treatment. Engineered endochondral cartilage grafts can integrate with host bone, undergo ossification, and heal critical-size long-bone defects in a mouse model. Level of Evidence. Animal study. Disclosure. A patent application on this technology has been filed

Background. Smoking has been associated with poor tissue oxygenation and vascularisation, predisposing smokers to a higher risk for postsurgical infections. The aim of this study was to estimate and compare the incidence of prosthetic joint infection (PJI) following primary total joint arthroplasty (TJA) according to smoking status. Methods. A prospective hospital-registry based cohort was used including all primary total knee and hip arthroplasties performed between 03/1996 and 12/2013 and following them until 06/2014. Smoking status at time of surgery was classified in never, former and current smoker. Incidence rates and incidence rate ratios (IRR) for PJI according to smoking status were assessed within the first year and over the whole study period. Adjusted IRRs were obtained using cox regression model. Adjustment was performed for the following baseline characteristics: age, sex, BMI, ASA score, diabetes, arthroplasty site (knee or hip) and surgery duration. Results. We included 8,559 TJAs, 3,361 knee and 5,198 hip arthroplasties. Mean age was 70 years, 61% were women, mean follow-up time was 77 months. 5,722 were never (group 1), 1,315 former (group 2) and 1,522 current (group 3) smokers. Over the study period, 108 PJI occurred. Incidence rates of infection within one year were for group 1, 2 and 3, respectively: 4.7, 10.1 and 10.9 cases/1000 person-year. Comparing ever- vs. never-smokers, the adjusted IRR was 1.84 (95% CI 1.05–3.2). Incidence rates for infection over the whole study period were 1.5, 3.1 and 2.7 cases/1000 person-years for group 1, 2 and 3, respectively. Adjusted IRR for ever- vs. never-smokers was 1.46 (95% CI 0.97–2.19). Conclusions. Smoking was associated with an about 1.5 times higher incidence rate of PJI following TJA. The difference was established already in the first year after surgery and remained thereafter. Level of Evidence. prospective registry based comparative cohort study (level II)