This study was an analysis of outcome for patients undergoing non-vascularised fibula reconstructions (segmental and hemicortical) following tumour resection at the extremities with respect to consolidation, hypertrophy at the graft-host junctions, complications and functional outcome. Therefore, we retrospectively evaluated 36 patients with bone tumours (malignant n = 15, benign n = 21) at the extremities (upper extremity n = 9, lower extremity n = 27) who were treated with non-vascularised fibula reconstructions. Primary union of the graft-host junctions was recorded in 94% of the patients (SR 87%, HR 100%) after a mean of 22 weeks, whereas non-union was seen in only 4% (SR 9%, HR 0%). The overall complication rate was 36%. There was a significant correlation between the development of mechanical complications (fracture, delayed-/non-union) and a defect size of ≥12 cm (p = 0.013), segmental defects (p = 0.013) and additional required treatment (p = 0.008).
The use of non-vascularised fibulae dates back to the beginning of the twentieth century [23], but this technique has increasingly faded into the background after Taylor’s first description of a vascularised fibula reconstruction [24], as vascularised bone grafts were said to have a higher potential for hypertrophy and/or remodelling [25,26,27].
However, reports in the literature have been controversial. Hypertrophy rates for vascularised fibula reconstructions vary between 37% and 90% [1, 12, 15, 21, 28]. Additionally, significant differences between the hypertrophy rates at the upper and lower extremities have been described. For example, Hsu et al. reported a hypertrophy rate of 75% at the lower limb but only one out of seven upper limbs experiencing hypertrophy, which was attributed to the lack of mechanical forces at the upper limb [12]. In contrast, a series by Hilven et al. describe hypertrophy rates of 100% at the upper extremities and 86% in the lower limbs, which were assumed to be associated with the longer period of load restriction at the lower extremities. Likewise, the ability of non-vascularised fibula reconstructions to undergo hypertrophy at the host site has been controversially discussed. It was shown that non-vascularised grafts are inferior with regard to integration, resistance to bacterial infection and hypertrophy compared to vascularised grafts [29]. Nevertheless, there is evidence in the literature that even non-vascularised bone grafts are capable of remodelling and integrating into the host bone [13, 18, 20, 30]. On one hand, this might be constituted as a creeping substitution with viable cells migrating from the well-perfused conjunction zone into the vascular graft. On the other hand, the integration of avascular grafts could be attributed to a periosteal hypertrophy leading to new bone formation around the graft and eventual bony integration of the graft in some cases [21]. Therefore, we evaluated the presence and extent of hypertrophy at the graft-host junctions. In our series, hypertrophy was recorded in 85% of the evaluated graft-host junctions, and 52% of these hypertrophies were significant (>20%). No statistically significant differences were found between the upper and the lower extremities. Furthermore, our results were comparable or even superior to those at the pelvis (67% hypertrophy) despite the good soft tissue coverage and blood supply in the pelvic region [18].
In our series, primary consolidation (defined as consolidation within 12 months after surgery) was seen in 94% (SR 85%, HR 100%) of the host-graft junctions, with delayed union in 2% (SR 4%, HR 0%) and non-union in only 4% (SR 9%, HR 0%) of the patients. This is markedly superior to reports by Enneking and Yadav, who described primary union rates (within 12 months) of 63% and 60% for non-vascularised fibula grafts at the extremities [31, 32]. Likewise, Schuh et al. reported a union rate (defined as trabecular bridging within 6 months after surgery) of 67% for non-vascularised fibulae (non-union rate of 33%) and 85% (non-union rate of 15%) for vascularised grafts [20]. Based on the criteria of Schuh et al., we would have achieved union in 70% of cases (SR 60%, HR 85%) [20]. However, in our series, more than half of the patients (58%) underwent hemicortical reconstructions (Fig. 2), which have presumably higher consolidation rates due to the larger contact area as well as the lower extent of soft tissue dissection [7]. Similar to our results, comparable studies on hemicortical reconstructions with auto- or allografts following tumour resection showed lower non-union rates of only 0–7% [7, 33,34,35,36].
The application of additional treatment modalities such as chemotherapy might be one factor that contributes to the prolonged time to union. Hariri et al. reported a mean union-time of 1.75 years using vascularised fibula grafts, but all patients received neo-/adjuvant chemotherapy [37]. In our series, a total of 6 patients received neo-/adjuvant therapy (Table 1), and three of these patients had a delayed union or non-union. Though the administration of neo-/adjuvant therapy did not significantly influence the union time (p = 0.58), a statistically significant correlation between additional treatment and delayed union or non-union (p = 0.003) as well as the development of mechanical complications (p = 0.006) was observed.
In our series, patients with plate fixation had a longer consolidation time (mean 27 weeks) compared to patients with screw/press-fit fixation (mean 20 weeks). However, this difference was not statistically significant (p = 0.234). We have the opinion that the differences in the consolidation time might have been a problem of insufficient primary stability rather than of the plate fixation itself. Independently from the fixation technique, the defect size was a main factor that influenced the union time as there was a highly significant correlation between defect size and union-time (p < 0.001, R2 = 0.35). Additionally, a defect size of 12 cm or greater (p = 0.013) as well as segmental reconstructions (p = 0.013) were statistically significant risk factors for suffering a mechanical complication. Thus, 4 of our 5 fatigue fractures occurred in single strut reconstructions and bone defects of 12 cm or more, 3 of which were segmental defects. This is in accordance with reports in the literature, where the superiority of vascularised fibulae over non-vascularised grafts was reported for bone defects longer than 12 cm as indicated by failure rates of 25% and 50%, respectively, [38]. The significantly lower mechanical complication rate of hemicortical reconstructions (p = 0.013) in our series might be explained—beside the above mentioned factors such as the limited extent of soft-tissue dissection and a greater contact surface between graft and host bone—by the preservation of cortical continuity [7]. Taking our own results into account, we therefore strongly recommend the use of vascularised fibula grafts for segmental bone defects of 12 cm or greater.
In patients who were included in our previous study [13], no complications were recorded during the last 8 years (after the end of the previous study). Thus, the overall complication rate was 36% (n = 13) in our patients with a mean follow up of 8.3 years (range 2.1–26.6 years), among this subset, 77% (n = 10) needed revision surgery. The revision rate in the study by Schuh et al. was slightly higher at 48% of non-vascularised fibula reconstructions and 73% of vascularised fibula grafts [20]. Interestingly, the use of vascularised fibula grafts, a short graft-length and a lower extremity were shown to be risk factors for revision [20]. Likewise, Hariri et al. reported on a mean re-operative rate of 2.02 per patient after reconstruction with vascularised fibula grafts and an infection rate of 16% [37]. The infection rate in our study was only 6% (n = 2), one of which was superficial. For alternative treatment options such as intercalary allograft reconstructions, the incidence of complications varies from 7.5–30% for infections and 30–63% for non-union or delayed unions [3, 6, 39]. In diaphyseal tumour endoprostheses, failure rates of up to 63% at 10 years have been published, and patients generally contend with a life-long risk for complications such as infection [3, 40].
One shortcoming of autologous fibula grafts is the risk of donor site complications such as peroneal nerve palsy, stress fractures or joint instability [18, 37]. In our series, the donor site morbidity was rather low (6%) compared to those reported for vascularised fibula grafts (7–36%) [15, 37, 41, 42]. Additionally, we believe that this risk is acceptable, at least in non-vascularised fibula reconstructions which offer the advantage of remodelling capabilities at the host site as well as a technically less demanding surgical technique. In contrast to vascularised fibula reconstructions, tibial stress fractures haven’t been reported for non-vascularised fibula grafts until now, which might be attributed to the remodelling capacity at the donor site. Thus, among the analysed cases, complete remodelling was seen in 44%, partial in 40% and no remodelling in only 16% of the cases. In accordance with Grzegirzewski et al., all patients younger than 12 years showed complete remodelling, and patients presenting no remodelling were all older than 29 years [43].
If insufficient bone stock is preserved at the distal fibula, the risk for instability and valgus deformity of the ankle is high, especially in children [44]. In our series, in which at least 8 cm was preserved at the distal tibia and 4–5 cm at the proximal end, neither instability nor deformity was recorded at the knee or ankle joint. Two patients (6%) suffered a transient peroneal nerve palsy which recovered completely over time. The functional outcome of our patients was appealing and comparable to alternative treatment options as indicated by the mean MSTS score of 86%. Likewise, MSTS scores between 78% and 92% have been reported for vascularised fibula grafts [20, 37] and between 84% and 90% for diaphyseal tumour endoprostheses [2, 45].
The study’s retrospective design, small sample size, and lack of control groups are its main limitations. Our cohort was somewhat heterogeneous with respect to tumour entities, resection technique (segmental vs. hemicortical), use of additional treatments, age and lesion localization. The study period duration was extensive, but there were no changes over time regarding the surgical technique for this procedure, and only 2 surgeons performed all the operations. However, to our knowledge, this is the largest series of non-vascularised fibula reconstructions that has been published to date.