Page 144 - Reconstructive Techniques in Musculoskeletal Tumor Surgery

Chapter 7
Reconstructions were located in the proximal tibia (n = 14, 37%), proximal humerus (n = 12, 32%), distal femur (n = 10, 26%) and distal radius (n = 2, 5%). A total of 32 patients (84%) had a segmental, and six (16%) a unicondylar reconstruction (four proximal tibial, two distal femoral). Median allograft length was 14 cm (IQR 10 to 17), and was greater for the proximal humerus (16.5 cm, IQR 13 to 19.5) than for other sites (12 cm, IQR 10 to 17). Allografts were xed to host bone using either one or more plate(s) (n = 28, 74%), an intramedullary nail (n = 8, 21%), screws only, or an intramedullary nail and plate (each; n = 1, 3%). In seven patients (18%), an allogeneic bular strut was used to reinforce the construct. A gastrocnemius ap was used in nine proximal tibial reconstructions (64%).
Systematic literature review
We performed a systematic search to identify papers on osteoarticular allograft reconstructions for musculoskeletal tumors. All clinical case series that reported on at least ve reconstructions, and were written in English, Dutch, German, French or Italian, were included. With search terms which are detailed in the supplementary material, we identi ed 244 unique titles in PubMed, EMBASE, Web of Science, CINAHL, Academic Search Premier and Science Direct. The leading author screened all titles and abstracts, extracted relevant data and critically appraised the included studies. The critical appraisal data were based on a previous systematic review on reconstructions for tumor resections35. Our review was registered with the international prospective register of systematic reviews (PROSPERO) database36 (identi er CRD42015026027).
Statistical analysis
Continuous variables were compared with Mann-Whitney U tests. Logistic regression analysis was used to assess the in uence of factors on the occurrence of complications in the retrospective study, and to compare the incidence of complications in the systematic review. Allograft survival was estimated with Kaplan-Meier curves with 95% con dence intervals (CI). We used SPSS v.21 software (IBM Corp., Armonk, New York), with the level of signi cance at a p-value < 0.05.
142