CORR Insights®: Causes and Frequencies of Reoperations After Endoprosthetic Reconstructions for Extremity Tumor Surgery: A Systematic Review

Where Are We Now?

Less than 50 years ago, amputation was the leading treatment option for patients with malignant bone tumors. But today, advances in imaging, (neo-)adjuvant treatment, surgical technique, and reconstructive options allow for limb salvage in the majority of patients without compromising patient survival and can—depending on the location—provide superior function and cosmesis.

Bone malignancies are relatively rare, and this leaves us with a paucity of high-quality evidence. But many case series have been published describing implant survival and the reasons for reoperation after limb-salvage surgery for bone malignancy. In the current study, Thornley and colleagues [11] summarized these studies in a comprehensive and thorough systematic review; however, one must be cognizant of the limitations inherent to the mainly retrospective single-center studies. While these studies were generally of low quality, it is important to recognize that standardized reporting in orthopaedic oncology has led to higher-quality studies in the past few decades. General guidelines, such as the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statements, Methodological index for nonrandomized studies (MINORS) criteria, and orthopaedic oncology specific recommendations, such as the Henderson classification contributed substantially to this improvement [3, 12]. The use of such guidelines decreases varied reporting across studies and facilitates bundling or pooling of data, which is valuable in rare diseases like primary bone tumors.

Where Do We Need To Go?

Generally, orthopaedic oncology studies report on implant survival, reasons for reoperation, limb salvage rate, tumor control, and complications. Fewer orthopaedic oncology studies report on functional outcome, of which most use the Musculoskeletal Tumor Society score, a clinician completed assessment of a patient's functional level [4]. Standardized reporting and clearly defined outcome, disease, and patient characteristics in both studies and institutional tumor registries would make it easier to combine data into larger and statistically powerful patient samples. In addition, prospective evaluations of pre and postoperative patient-reported outcome measures (PROMs) are needed to better understand our patients’ symptoms and limitations [1, 9]. These evaluations will also enable us to study how various aspects of treatment and adverse events influence PROMs. Patient-reported outcome measures are preferable to clinician-assessed measures, as the latter can be biased when judging symptoms and limitations [6, 7].

Because randomized controlled trials are extremely difficult and costly to perform in orthopaedic oncology, priority should be given to collecting standardized, prospective, large-sample patient datasets. Given the rarity and heterogeneity of bone malignancies, multicenter collaboration is needed to drive advances in evidence-based care. Consensus should be reached in the international orthopaedic oncology community, as well as with affiliated researchers, on how to measure implant survival, modes of failure, tumor control, physical function, and quality of life. We also need to determine which patient and disease characteristics should be included in these types of studies and at what time points.

How Do We Get There?

Perhaps we can utilize the same questionnaires that have already been used to assess validated PROMs for patients with sarcoma undergoing surgical treatment. Delphi studies can be used to reach consensus among a large panel of experts, preferably from multiple specialties [2]. For example, Schneider and colleagues [8] used a Delphi approach to determine top research priorities in orthopaedic oncology among members of the Musculoskeletal Tumor Society.

Collecting prospective observational data on an easy-to-use secure online platform linked to hospital medical records and to surveys sent to patients can help us build and maintain a multi-institutional registry. Machine-learning models [5, 10]—the ability of a computer to learn from a large set of examples without being explicitly programmed (spam email filtering, for example)—can theoretically be fed by this constantly updating registry and could provide instant feedback in terms of outcome probabilities with corresponding confidence intervals that can guide future decision making. But the development of machine-learning prediction models and their validity is something we need to research further [10]. We can minimize the burden on participating patients and orthopaedic oncologists by decreasing the number of items that need to be completed for registries. Therefore, consensus among experts on what data to collect is important. In addition, part of the registry could be completed automatically from the hospital medical record or could be completed by local research or registry coordinators. Exploring how to overcome barriers to funding, as well as international multi-institutional collaboration could be a valuable first step towards achieving the aforementioned goals.