En-bloc resection was thought as the standard treatment method since Enneking staging system was proposed, but former reports had proven that geographic resection using multiplanar osteotomies was an effective treatment method for a bone sarcoma [1, 20]. The preservation of host bone or juxta-articular bone with reconstructions could help to achieve better limb functions in the long-term follow up. However, the inaccurate preoperative planning and tumor resection may not achieve the safe surgical margins and intricate the local recurrence of bone tumor [21].
The use of computer navigation technology in bone tumor resections had greatly improve the surgical margins in some studies. Jeys et.al. Reported the computer navigation can help reduce the risk of an intralesional margin and reduce the rate of local recurrence (13%) in patients with a tumour of the pelvis and sacrum, which clear bone resection margins were achieved in all cases (31/31) [18]. And Gerbers et.al. reported the preliminary resection accuracy result of computer navigation surgery in 4 cases with an average gap of 0.9 (0–5.4) mm [22–25].
In this study, we aim to address the following three questions about computer navigation assisted geographic resections: (1) What is the resection accuracy difference between the computer-navigation assisted and free-hand geographic resection? (2) Does computer navigation technology affect the oncological and functional outcomes of primary bone tumor in lower limbs? (3) Is the comlications acceptable for geographic resection combined with allograft bone reconstruction in low-grade maligant and benign bone tumors? To our knowledge, this study represents the largest reported sample size on computer-navigation assisted geographic resections and allograft reconstructions in English literatures.
Accuracy comparasion of geographic resection
The current study showed the average maximum resection errors in computer-navigation assisted geographic surgery were 1.760 ± 1.109 mm, which was slightly better than former studies’ results [12, 15, 26–30]. In a cadaveric study, Bsoma et.al. reported the accuracy comparasion of the free-hand group, computer navigation assistance (CAS), patient specific guidance (PSG) and CAS + PSG group. The mean resection errors were 9.2 ± 3.3 mm in the free-hand group while the CAS group, PSG group and CAS + PSG group achieved an mean resection error of 3.6 ± 3.3 mm, 1.9 ± 1.1 mm and 2.0 ± 1.0 mm, respectively. However, the cadaveric study didnot include the adequate margin of bone and soft tissue tumor which should be thought as the first consideration [15]. Wong et.al. reported 4 cases with the mean manximum errors of 1.61 ± 0.35 mm by using the computer navigation assistance and patient specific guidance placement, which improved the acurracy of resection errors. However, the small sample size and limited indications (juxta-articular bones) may influence the results and restrict the further use in bone tumor surgeries [29].
Moreover, the robot-assisted surgery has been becoming an promising treatment methods for loacting and resection which the robot arm can stably move and reduce the micromotion of free-hand surgery. Therefore, the preoperative plan could be precisely implemented and may futher improve the resection accuracy and the resected margins of bone tumors, especially those in inregular bones [30].
Oncological and functional outcomes
Surgical margin is the important indepedent risk factor to the overall survival and local recurrence reported in former studies [2, 18, 29, 31].Clear margins were achieved by computer-navigation assisted surgery in all 25 cases according to the evaluation of postoperative specimens by pathologists in current study. However, two cases (2 giant celle tumors of bone) were found positive margins in the free-hand group which prompted the importance of navigaition use in geographic resection for giant cell tumor of bone. Finally, one case underwent amputation due to local recurrence and no evidence of disease until the latest follow-up.
Local recurrence cases occuered in three patients (1 periosteal osteosarcoma, 1 common chondrosarcoma and 1 giant cell tumor of bone ) with intralesional and marginal resection. Even some studies reported the limited resections would result in higher local recurrence rate in the maligant bone tumors[3, 30–33], we thought the preservation of host bone and excellent long-term functional outcomes overweigh the possible higher recurrence rates which was not seen in our results. Chen et.al. reported the hemicortical resection and biological reconstructions of high-grade osteosarcomas, which obtained clear margins in all 6 cases and no evidence of loacal recurrence and distant metastasis were found at a median follow-up of 50.5 months [34]. In current study, no high-grade bone sarcoma cases were planned to accept geographic resections, however 3 cases were diagnosed as high-grade bone sarcomas (1 parosteal osteosarcoma, 1 high-grade surface osteosarcoma, 1 dedifferentiated chondeosarcoma) according to the specimen examinations. Finally, even all 3 patients accepted postoperative chemotherapies, one patient (high-grade surface osteosarcoma) deceased due to the pulmonary metastasis. Then, we donnot recommend the regular use of the limited resection and further studies are needed to testify the advantages of this technique in high-grade bone sarcomas.
Complications
The most common complications were nonunion of allograft bones (8%) in this study, which were reported as rates of 7–26% in former studies [3, 4, 21]. Although the free-hand group (3/26; 11.5%) had a trend of higher rates of ununion complications than the navigation group (1/25; 4%), there was no statistical difference in ununion rates between the two groups. Even though autograft use may improve the unnion rates[35–36], while the specific location of our cases, which inclueded the distal femur and femoral chondyles, may not be suitable for the incorporation of autograft bones.
Fractures had been reported as the most common complication in former study [4], while there were also some studies reported no fractures which didnot describe the resection extent of the cortical bone [37]. Better fit of the allografts and host bones had been reported to reduce the incidence rate of fractures [38], the current study improved the max gap error to get better match between the allograft and host bone which may be the possible factor of the relatively low incidence rates of fractures (4%) in this study.
Infection rate was 4% in this study, which had no difference between the groups. One patient accepted the endoprosthetic removeal and biological reconstruction as a treatment. The former studies reported infection rates ranging from 0–10% [4, 39], tibial diaphysis had a higher risk of infection, while the lack of soft tissue coverage may be a possible reason. In current study, we did not use the muscle flap transfer techniques to cover the allograft and endoprosthesis, because the limited dissection of geographic resections helped to preserve the soft tissue around the reconstruction locations.
Study limitations
There are also some limitations that should be adressed in this study. Firstly, the retrospective, non-randomized nature of this study inevitably brought seletive and memorial bias. What’s more, the limited sample size also restricted the statistical analysis stability in current study. Secondly, more precise resection error evaluation methods should be used to acurrately analyze the difference between the computer-navigation assisted group and free-hand group. The computer assisted analysis methods may be a considerable choice for the resection error evaluation [26, 40]. Thirdly, Further follow-up is necessary for detecting possible complications of the allografts reconstruction which include allograft fracture, host bone fracture and bone absorption.