Aims. Osteoarticular reconstruction of the distal femur in childhood has the advantage of preserving the tibial physis. However, due to the small size of the distal femur, matching the host bone with an osteoarticular allograft is challenging. In this study, we compared the outcomes and complications of a resurfaced allograft-prosthesis composite (rAPC) with those of an osteoarticular allograft to reconstruct the distal femur in children. Methods. A retrospective analysis of 33 skeletally immature children with a malignant tumour of the distal femur, who underwent resection and reconstruction with a rAPC (n = 15) or osteoarticular allograft (n = 18), was conducted. The median age of the patients was ten years (interquartile range (IQR) 9 to 11) in the osteoarticular allograft group and nine years (IQR 8 to 10) in the rAPC group (p = 0.781). The median follow-up of the patients was seven years (IQR 4 to 8) in the osteoarticular allograft group and six years (IQR 3 to 7) in the rAPC group (p = 0.483). Limb function was evaluated using the Musculoskeletal Tumor Society (MSTS) score. Results. At final follow-up, the knee was unstable in 9/18 patients (50%) in the osteoarticular allograft group and 2/15 patients (13%) in the rAPC group (p = 0.026). The median range of motion (ROM) of the knee was 117° (IQR 115° to 120°) in the osteoarticular allograft group and 100° (IQR 95° to 105°) in the rAPC group (p < 0.001). The median MSTS score was 25 (IQR 23 to 26) in the osteoarticular allograft group and 28 (IQR 26 to 29) in the rAPC group (p = 0.007). Osteoarthritic change was detected in 11/18 patients (61%) in the osteoarticular allograft group and in 4/15 (26%) patients in the rAPC group (p = 0.048). Conclusion. In our series, a resurfaced allograft-prosthesis composite provided better knee stability and function, with a lower rate of osteoarthritis; an osteoarticular allograft was associated with better knee ROM. Cite this article: Bone Joint J 2022;104-B(10):1174–1179

Aims. To clarify the mid-term results of transposition osteotomy of the acetabulum (TOA), a type of spherical periacetabular osteotomy, combined with structural allograft bone grafting for severe hip dysplasia. Methods. We reviewed patients with severe hip dysplasia, defined as Severin IVb or V (lateral centre-edge angle (LCEA) < 0°), who underwent TOA with a structural bone allograft between 1998 and 2019. A medical chart review was conducted to extract demographic data, complications related to the osteotomy, and modified Harris Hip Score (mHHS). Radiological parameters of hip dysplasia were measured on pre- and postoperative radiographs. The cumulative probability of TOA failure (progression to Tönnis grade 3 or conversion to total hip arthroplasty) was estimated using the Kaplan–Meier product-limited method, and a multivariate Cox proportional hazard model was used to identify predictors for failure. Results. A total of 64 patients (76 hips) were included in this study. The median follow-up period was ten years (interquartile range (IQR) five to 14). The median mHHS improved from 67 (IQR 56 to 80) preoperatively to 96 (IQR 85 to 97) at the latest follow-up (p < 0.001). The radiological parameters improved postoperatively (p < 0.001), with the resulting parameters falling within the normal range in 42% to 95% of hips. The survival rate was 95% at ten years and 80% at 15 years. Preoperative Tönnis grade 2 was an independent risk factor for TOA failure. Conclusion. Our findings suggest that TOA with structural bone allografting is a viable surgical option for correcting severely dysplastic acetabulum in adolescents and young adults without advanced osteoarthritis, with favourable mid-term outcomes. Cite this article: Bone Joint J 2023;105-B(7):743–750

Aims. Tibial tubercle osteotomy (TTO) facilitates surgical exposure and protects the extensor mechanism during revision total knee arthroplasty (rTKA). The purpose of this study was to determine the rates of bony union, complications, and reoperations following TTO during rTKA, to assess the functional outcomes of rTKA with TTO at two years’ minimum follow-up, and to identify the risk factors of failure. Methods. Between January 2010 and September 2020, 695 rTKAs were performed and data were entered into a prospective database. Inclusion criteria were rTKAs with concomitant TTO, without extensor mechanism allograft, and a minimum of two years’ follow-up. A total of 135 rTKAs were included, with a mean age of 65 years (SD 9.0) and a mean BMI of 29.8 kg/m. 2. (SD 5.7). The most frequent indications for revision were infection (50%; 68/135), aseptic loosening (25%; 34/135), and stiffness (13%; 18/135). Patients had standardized follow-up at six weeks, three months, six months, and annually thereafter. Complications and revisions were evaluated at the last follow-up. Functional outcomes were assessed using the Knee Society Score (KSS) and range of motion. Results. The mean follow-up was 51 months (SD 26; 24 to 121). Bony union was confirmed in 95% of patients (128/135) at a mean of 3.4 months (SD 2.7). The complication rate was 15% (20/135), consisting of nine tibial tubercle fracture displacements (6.7%), seven nonunions (5%), two delayed unions, one tibial fracture, and one wound dehiscence. Seven patients (5%) required eight revision procedures (6%): three bone grafts, three osteosyntheses, one extensor mechanism allograft, and one wound revision. The functional scores and flexion were significantly improved after surgery: mean KSS knee, 48.8 (SD 17) vs 79.6 (SD 20; p < 0.001); mean KSS function, 37.6 (SD 21) vs 70.2 (SD 30; p < 0.001); mean flexion, 81.5° (SD 33°) vs 93° (SD 29°; p = 0.004). Overall, 98% of patients (n = 132) had no extension deficit. The use of hinge implants was a significant risk factor for tibial tubercle fracture (p = 0.011). Conclusion. TTO during rTKA was an efficient procedure to improve knee exposure with a high union rate, but had significant specific complications. Functional outcomes were improved at mid term. Cite this article: Bone Joint J 2023;105-B(10):1078–1085

Aims. Long-term outcomes following the use of human dermal allografts in the treatment of symptomatic irreparable rotator cuff tears are not known. The aim of this study was to evaluate these outcomes, and to investigate whether this would be a good form of treatment in young patients in whom a reverse shoulder arthroplasty should ideally be avoided. Methods. This prospective study included 47 shoulders in 45 patients who underwent an open reconstruction of the rotator cuff using an interposition GraftJacket allograft to bridge irreparable cuff tears, between January 2007 and November 2011. The Oxford Shoulder Score (OSS), pain score, and range of motion (ROM) were recorded preoperatively and at one year and a mean of 9.1 years (7.0 to 12.5) postoperatively. Results. There was significant improvement in the mean OSS from 24.7 (SD 5.4) preoperatively to 42.0 (SD 6.3) at one year, and this improvement was maintained at 9.1 years (p < 0.001), with a score of 42.8 (SD 6.8). Similar significant improvements in the pain score were seen and maintained at the final follow-up from 6.1 (SD 1.6) to 2.1 (SD 2.3) (p < 0.001). There were also significant improvements in the ROM of the shoulder, and patient satisfaction was high. Conclusion. The use of an interposition human dermal allograft in patients with an irreparable rotator cuff tear leads to good outcomes that are maintained at a mean of nine years postoperatively. Cite this article: Bone Joint J 2022;104-B(1):91–96

Aims. Giant cell tumours (GCTs) of the proximal femur are rare, and there is no consensus about the best method of filling the defect left by curettage. In this study, we compared the outcome of using a fibular strut allograft and bone cement to reconstruct the bone defect after extended curettage of a GCT of the proximal femur. Methods. In a retrospective study, we reviewed 26 patients with a GCT of the proximal femur in whom the bone defect had been filled with either a fibular strut allograft (n = 12) or bone cement (n = 14). Their demographic details and oncological and nononcological complications were retrieved from their medical records. Limb function was assessed using the Musculoskeletal Tumor Society (MSTS) score. Results. Mean follow-up was 116 months (SD 59.2; 48 to 240) for the fibular strut allograft group and 113 months (SD 43.7; 60 to 192) for the bone cement group (p = 0.391). The rate of recurrence was not significantly different between the two groups (25% vs 21.4%). The rate of nononcological complications was 16.7% in the strut allograft group and 42.8% in the bone cement group. Degenerative joint disease was the most frequent nononcological complication in the cement group. The mean MSTS score of the patients was 92.4% (SD 11.5%; 73.3% to 100.0%) in the fibular strut allograft group and 74.2% (SD 10.5%; 66.7% to 96.7%) in the bone cement group (p < 0.001). Conclusion. Given the similar rate of recurrence and a lower rate of nononcological complications, fibular strut grafting could be recommended as a method of reconstructing the bone defect left by curettage of a GCT of the proximal femur. Cite this article: Bone Joint J 2022;104-B(2):297–301

Periprosthetic joint infection represents a devastating complication after total elbow arthroplasty. Several measures can be implemented before, during, and after surgery to decrease infection rates, which exceed 5%. Debridement with antibiotics and implant retention has been reported to be successful in less than one-third of acute infections, but still plays a role. For elbows with well-fixed implants, staged retention seems to be equally successful as the more commonly performed two-stage reimplantation, both with a success rate of 70% to 80%. Permanent resection or even amputation are occasionally considered. Not uncommonly, a second-stage reimplantation requires complex reconstruction of the skeleton with allografts, and the extensor mechanism may also be deficient. Further developments are needed to improve our management of infection after elbow arthroplasty. Cite this article: Bone Joint J 2024;106-B(11):1321–1326

Aims. The aim of this study was to evaluate the clinical and radiological outcomes of locking plate fixation, with and without an associated fibular strut allograft, for the treatment of displaced proximal humeral fractures in elderly osteoporotic patients. Patients and Methods. We undertook a retrospective comparison of two methods of fixation, using a locking plate without an associated fibular strut allograft (LP group) and with a fibular allograft (FA group) for the treatment of these fractures. The outcome was assessed for 52 patients in the LP group and 45 in the FA group, with a mean age of 74.3 years (52 to 89), at a mean follow-up of 14.2 months (12 to 19). The clinical results were evaluated using a visual analogue scale (VAS) score for pain, the Constant score, the American Shoulder and Elbow Surgeons (ASES) score, and the range of movement. Radiological results were evaluated using the neck-shaft angle (NSA) and humeral head height (HHH). Results. The mean forward elevation in the LP and FA groups was 125.3° (. sd. 21.4) and 148.9° (. sd. 19.8), respectively (p = 0.042), while other clinical factors showed no statistically significant differences between the groups. The changes in NSA and HHH immediately after the operation and at final follow-up were significantly better in the FA group than in the LP group (p = 0.015 and p = 0.021, respectively). Conclusion. For comminuted proximal humeral fractures in osteoporotic patients, locking plate fixation with a fibular strut allograft shows satisfactory short-term results with respect to humeral head support and maintenance of reduction, and may reduce the incidence of complications associated with fixation using a locking plate alone. Cite this article: Bone Joint J 2019;101-B:260–265

Cartilage defects of the hip cause significant pain and may lead to arthritic changes that necessitate hip replacement. We propose the use of fresh osteochondral allografts as an option for the treatment of such defects in young patients. Here we present the results of fresh osteochondral allografts for cartilage defects in 17 patients in a prospective study. The underlying diagnoses for the cartilage defects were osteochondritis dissecans in eight and avascular necrosis in six. Two had Legg-Calve-Perthes and one a femoral head fracture. Pre-operatively, an MRI was used to determine the size of the cartilage defect and the femoral head diameter. All patients underwent surgical hip dislocation with a trochanteric slide osteotomy for placement of the allograft. The mean age at surgery was 25.9 years (17 to 44) and mean follow-up was 41.6 months (3 to 74). The mean Harris hip score was significantly better after surgery (p < 0.01) and 13 patients had fair to good outcomes. One patient required a repeat allograft, one patient underwent hip replacement and two patients are awaiting hip replacement. Fresh osteochondral allograft is a reasonable treatment option for hip cartilage defects in young patients. Cite this article: Bone Joint J 2014;96-B(11 Supple A):11–16

Bone allografts can be used in any kind of surgery involving bone from minor defects to major bone loss after tumour resection. This review describes the various types of bone grafts and the current knowledge on bone allografts, from procurement and preparation to implantation. The surgical conditions for optimising the incorporation of bone are outlined, and surgeon expectations from a bone allograft discussed

Aims. The management of acetabular defects at the time of revision hip arthroplasty surgery is a challenge. This study presents the results of a long-term follow-up study of the use of irradiated allograft bone in acetabular reconstruction. Patients and Methods. Between 1990 and 2000, 123 hips in 110 patients underwent acetabular reconstruction for aseptic loosening, using impaction bone grafting with frozen, irradiated, and morsellized femoral heads and a cemented acetabular component. A total of 55 men and 55 women with a mean age of 64.3 years (26 to 97) at the time of revision surgery are included in this study. Results. At a mean follow-up of 16.9 years, there had been 23 revisions (18.7%), including ten for infection, eight for aseptic loosening, and three for dislocation. Of the 66 surviving hips (58 patients) that could be reassessed, 50 hips (42 patients; 75.6%) were still functioning satisfactorily. Union of the graft had occurred in all hips with a surviving implant. Survival analysis for all indications was 80.6% at 15 years (55 patients at risk, 95% confidence interval (CI) 71.1 to 87.2) and 73.7% at 20 years (eight patients at risk, 95% CI 61.6 to 82.5). Conclusion. Acetabular reconstruction using frozen, irradiated, and morsellized allograft bone and a cemented acetabular component is an effective method of treatment. It gives satisfactory long-term results and is comparable to other types of reconstruction. Cite this article: Bone Joint J 2018;100-B:1449–54

Aim. There is insufficient evidence to support bony reconstruction of the pubis after a type III internal hemipelvectomy (resection of all or part of the pubis). In this study, we compared surgical complications, postoperative pain, and functional outcome in a series of patients who had undergone a type III internal hemipelvectomy with or without bony reconstruction. Methods. In a retrospective cohort study, 32 patients who had undergone a type III hemipelvectomy with or without allograft reconstruction (n = 15 and n = 17, respectively) were reviewed. The mean follow-up was 6.7 years (SD 3.8) for patients in the reconstruction group and 6.1 years (SD 4.0) for patients in the non-reconstruction group. Functional outcome was evaluated using the Musculoskeletal Tumor Society (MSTS) scoring system and the level of postoperative pain with a visual analogue scale (VAS). Results. The mean MSTS score of the patients was significantly better in patients after reconstruction (26 (SD 1.7) vs 22.7 (SD 2.0); p < 0.001). The mean visual analogue scale score for pain was significantly less in the reconstruction group (2.1 (SD 2) vs 4.2 (SD 2.2); p = 0.016). One infection occurred in each group. Bladder herniation occurred in three patients (17.6%) in the non-reconstruction group but none in the reconstruction group. Five patients (29.4%) in the non-reconstruction group and one (7%) in the reconstruction group had a limp. Graft displacement occurred in two patients in the reconstruction group. Conclusion. We recommend reconstruction of the bony defect after a type III hemipelvectomy: it gives a better functional result, less postoperative pain, and fewer late surgical complications. Cite this article: Bone Joint J 2021;103-B(6):1155–1159

Allograft bone is widely used in orthopaedic surgery, but peri-operative infection of the graft remains a common and disastrous complication. The efficacy of systemic prophylactic antibiotics is unproven, and since the graft is avascular it is likely that levels of antibiotic in the graft are low. Using an electrical potential to accelerate diffusion of antibiotics into allograft bone, high levels were achieved in specimens of both sheep and human allograft. In human bone these ranged from 187.1 mg/kg in endosteal (. sd. 15.7) to 124.6 (. sd. 46.2) in periosteal bone for gentamicin and 31.9 (. sd. 8.9) in endosteal and 2.9 (. sd. 1.1) in periosteal bone for flucloxacillin. The antibiotics remained active against bacteria in vitro after iontophoresis and continued to elute from the allograft for up to two weeks. Structural allograft can be supplemented directly with antibiotics using iontophoresis. The technique is simple and inexpensive and offers a potential means of reducing the rate of peri-operative infection in allograft surgery. Iontophoresis into allograft bone may also be applicable to other therapeutic compounds

Aims. Intercalary allografts following resection of a primary diaphyseal tumour have high rates of complications and failures. At our institution intercalary allografts are augmented with intramedullary cement and fixed using compression plating. Our aim was to evaluate their long-term outcomes. Patients and Methods. A total of 46 patients underwent reconstruction with an intercalary allograft between 1989 and 2014. The patients had a mean age of 32.8 years (14 to 77). The most common diagnoses were osteosarcoma (n = 16) and chondrosarcoma (n = 9). The location of the tumours was in the femur in 21, the tibia in 16 and the humerus in nine. Function was assessed using the Musculoskeletal Tumor Society (MSTS) scoring system and the Toronto Extremity Salvage Score (TESS). The survival of the graft and the overall survival were assessed using the Kaplan-Meier method. Results. The median follow-up was 92 months (4 to 288). The mean MSTS 87 score was 29.1 (19 to 35), the mean MSTS 93 score was 82.2 (50 to 100) and the mean TESS score was 81.2 (43 to 100). Overall survival of the allograft was 84.8%. A total of 15 patients (33%) had a complication. Five allografts were revised for complications and one for local recurrence. Conclusion. Intercalary allografts augmented with intramedullary cement and compression plate fixation provide a reliable and durable method of reconstruction after the excision of a primary diaphyseal bone tumour, with high levels of function and satisfaction. . Cite this article: Bone Joint J 2017;99-B:973–8

Structural allografts may be used to manage uncontained bone defects in revision total knee replacement (TKR). However, the availability of cadaver grafts is limited in some areas of Asia. The aim of this study was to evaluate the mid-term outcome of the use of femoral head allografts for the reconstruction of uncontained defects in revision TKR, focusing on complications related to the graft. We retrospectively reviewed 28 patients (30 TKRs) with Anderson Orthopaedic Research Institute (AORI) type 3 bone defects, who underwent revision using femoral head allografts and stemmed components. The mean number of femoral heads used was 1.7 (1 to 3). The allograft–host junctions were packed with cancellous autograft. At a mean follow-up of 76 months (38 to 136) the mean American Knee Society knee score improved from 37.2 (17 to 60) pre-operatively to 90 (83 to 100) (p < 0.001). The mean function score improved from 26.5 (0 to 50) pre-operatively to 81 (60 to 100) (p < 0.001). All the grafts healed to the host bone. The mean time to healing of the graft was 6.6 months (4 to 16). There have been no complications of collapse of the graft, nonunion, infection or implant loosening. No revision surgery was required. The use of femoral head allografts in conjunction with a stemmed component and autogenous bone graft in revision TKR in patients with uncontained bone defects resulted in a high rate of healing of the graft with minimal complications and a satisfactory outcome. Longer follow-up is needed to observe the evolution of the graft. Cite this article: Bone Joint J 2013;95-B:643–8

The outcome of tibial allograft reconstruction after resection of a tumour is inconsistent and has a high rate of failure. There are few reports on the use of tibial allografts in children with open growth plates. We performed 21 allograft reconstructions (16 osteoarticular, five intercalary) in 19 consecutive patients between seven and 17 years of age. Two had Ewing’s sarcoma, one an adamantinoma and 16 osteosarcoma, one with multifocal disease. Five patients have died; the other 14 were free from disease at the time of follow-up. Six surviving patients (eight allograft reconstructions) continue to have good or excellent function at a mean of 59 months (14 to 132). One patient has poor function at 31 months. The other seven patients have a good or excellent function after additional procedures including exchange of the allograft and resurfacing or revision to an endoprosthesis at a mean of 101 months (43 to 198). The additional operations were performed at a mean of 47 months (20 to 84) after the first reconstruction. With the use of allograft reconstruction in growing children, joints and growth plates may be preserved, at least partially. Although our results remain inconsistent, tibial allograft reconstruction in selected patients may restore complete and durable function of the limb

Revision arthroplasty after infection can often be complicated by both extensive bone loss and a relatively high rate of re-infection. Using allograft to address the bone loss in such patients is controversial because of the perceived risk of bacterial infection from the use of avascular graft material. We describe 12 two-stage revisions for infection in which segmental allografts were loaded with antibiotics using iontophoresis, a technique using an electrical potential to drive ionised antibiotics into cortical bone. Iontophoresis produced high levels of antibiotic in the allograft, which eluted into the surrounding tissues. We postulate that this offers protection from infection in the high-risk peri-operative period. None of the 12 patients who had two-stage revision with iontophoresed allografts had further infection after a mean period of 47 months (14 to 78)

The use of allografts for the treatment of bone tumours in children is limited by nonunion and the difficulty of finding a suitable graft. Furthermore, appositional growth can’t be expected of an allograft. We used an overlapping allograft in 11 children, with a mean age of ten years (4 to 15), with a mean follow-up of 24.1 months (20 to 33). There were five intercalary and six intra-articular resections, and the tumours were in the femur in six cases and the humerus in five. Rates of union, times to union, remodelling patterns and allograft-associated complications were evaluated. No allograft was removed due to a complication. Of the 16 junctional sites, 15 (94%) showed union at a mean of 3.1 months (2 to 5). Remodelling between host and allograft was seen at 14 junctions at a mean of five months (4 to 7). The mean Musculoskeletal Tumor Society score was 26.5 of 30 (88.3%). One case of nonunion and another with screw protrusion required re-operation. Overlapping allografts have the potential to shorten time to union, decrease rates of nonunion and have positive appositional growth effect

We retrospectively reviewed 40 hips in 36 patients who had undergone acetabular reconstruction using a titanium Kerboull-type acetabular reinforcement device with bone allografts between May 2001 and April 2006. Impacted bone allografts were used for the management of American Academy of Orthopaedic Surgeons Type II defects in 17 hips, and bulk bone allografts together with impacted allografts were used for the management of Type III defects in 23 hips. A total of five hips showed radiological failure at a mean follow-up of 6.7 years (4.5 to 9.3), two of which were infected. The mean pre-operative Merle d’Aubigné score was 10 (5 to 15) vs 13.6 (9 to 18) at the latest follow-up. The Kaplan-Meier survival rate at ten years, calculated using radiological failure or revision of the acetabular component for any reason as the endpoint, was 87% (95% confidence interval 76.3 to 97.7). A separate experimental analysis of the mechanical properties of the device and the load-displacement properties of bone grafts showed that a structurally hard allograft resected from femoral heads of patients with osteoarthritis should be preferentially used in any type of defect. If impacted bone allografts were used, a bone graft thickness of < 25 mm was acceptable in Type II defects. This clinical study indicates that revision total hip replacement using the Kerboull-type acetabular reinforcement device with bone allografts yielded satisfactory mid-term results

Aims. The aim of this study was to report the outcome of femoral condylar fresh osteochondral allografts (FOCA) with concomitant realignment osteotomy with a focus on graft survivorship, complications, reoperation, and function. Patients and Methods. We identified 60 patients (16 women, 44 men) who underwent unipolar femoral condylar FOCA with concomitant realignment between 1972 and 2012. The mean age of the patients was 28.9 years (10 to 62) and the mean follow-up was 11.4 years (2 to 35). Failure was defined as conversion to total knee arthroplasty, revision allograft, or graft removal. Clinical outcome was evaluated using the modified Hospital for Special Surgery (mHSS) score. Results. A total of 14 grafts (23.3%) failed at a mean of 8.6 years (1.4 to 20.1). Graft survivorship was 87.3% (95% confidence interval (CI) 79.0 to 96.6), 85.0% (95% CI 75.8 to 95.3), 74.8% (95% CI 62.2 to 90.0), 65.2% (95% CI 49.9 to 85.2), and 59.8% (95% CI 43.5 to 82.1) at five, ten, 15, 20, and 25 years, respectively. A total of 23 patients (38.3%) developed complications, and 26 (43.3%) had a further operation. Persistent postoperative malalignment occurred more frequently in failed grafts (28.6% vs 4.3%; p = 0.023), and was a risk factor for graft failure (hazard ratio 6.55; 95% CI 1.61 27.71; p = 0.009). The mean mHSS score improved from 74.1 (40 to 91) preoperatively to 89.0 (66 to 100) at final follow-up (p < 0.001). Conclusion. Femoral condylar FOCA with concomitant realignment osteotomy provides excellent long-term graft survival and reliable functional improvement. Persistent malalignment may increase the risk for graft failure

Iontophoresis is a novel technique which may be used to facilitate the movement of antibiotics into the substance of bone using an electrical potential applied externally. We have examined the rate of early infection in allografts following application of this technique in clinical practice. A total of 31 patients undergoing revision arthroplasty or surgery for limb salvage received 34 iontophoresed sequential allografts, of which 26 survived for a minimum of two years. The mean serum antibiotic levels after operation were low (gentamicin 0.37 mg/l (0.2 to 0.5); flucloxacillin 1 mg/l (0 to 1) and the levels in the drains were high (gentamicin 40 mg/l (2.5 to 131); flucloxacillin 17 mg/l (1 to 43). There were no early deep infections. Two late infections were presumed to be haemotogenous; 28 of the 34 allografts were retained. In 12 patients with pre-existing proven infection further infection has not occurred at a mean follow-up of 51 months (24 to 82)