Editorial Comment: Selected Proceedings From the 2023 Musculoskeletal Tumor Society Meeting

H. Thomas Temple MD

The selected proceedings from the 2023 Musculoskeletal Tumor Society (MSTS) Meeting in Clinical Orthopaedics and Related Research® highlight some exciting topics in research and clinical orthopaedic oncology surgery. I congratulate the MSTS program committee along with the authors of these papers for identifying some of the most urgent clinical topics, which range from basic science investigations to studies on evolving information technologies that will improve our ability to gather and interpret large data like never before.

The use of serum samples to detect tumor DNA is both provocative and daunting, especially since sarcomas are individually and collectively heterogeneous. In Anderson et al. [1], circulating tumor DNA (ctDNA) was detected in plasma samples from patients with sarcomas when subjected to low-passage whole genome sequencing (LP-WGS). The technique was less effective for samples from patients with sarcomas that were characterized by a translocation like synoviosarcoma or clear cell sarcoma. Because sarcomas have structural changes to their chromosomes—including gains and losses of part or entire chromosomes, known as copy number alterations (CNAs)—detecting ctDNA in serum could serve as a target for a liquid biopsy. What will make interpretation difficult, though, is that the presence, type, and quantity of ctDNA are affected by when the specimen was taken (time of diagnosis, during or after treatment, and surveillance) and whether the tumor is primary or metastatic. Nonetheless, these findings set the stage for further work on liquid biopsies for sarcomas.

Periprosthetic joint infection (PJI) is a common cause of revision following reconstruction after tumor removal [3]. Advancements in PJI treatment for routine hip and knee arthroplasties have resulted in eradication rates above 90% for conventional implants treated with two-stage revision [7]. In contrast, similar improvements have not been observed in patients with megaprosthesis PJI, with eradication rates ranging from 50% [8] to 75% [2]. In this month’s CORR®, McChesney et al. [5] show that a one-stage procedure—debridement of the articulation and antibiotic treatment with some implant retention (DAIR)—was not as effective as a two-stage procedure in eradicating infection for megaprosthetic PJI. Of course, results of PJI treatment have been shown to be influenced by many other mitigating factors, such as type of organism, time interval between diagnosis of infection and intervention, the presence of a draining sinus, and other host factors [9]. Because the overall number of megaprosthetic reconstructions after tumor resection is relatively small, follow-up studies about PJI after these procedures will probably need to be multicenter efforts. This is an opportunity for the MSTS to encourage its members to work together on a study aimed at making definitive recommendations and clinical guidelines for patients with infected megaprostheses.

Natural language processing (NLP) gathers, processes, and organizes large amounts of free-text data more efficiently than humans [10]. One of many applications of artificial intelligence (AI) in orthopaedic oncology is the identification of patient- and disease-specific data in building datasets. In these pages of CORR®, Lindsay et al. [4] show how de novo NLP was superior to ICD-9 codes in identifying patients with myxofibrosarcoma in the National Veterans Administration database. The NLP identified 92% of diagnoses, whereas 25% of diagnoses were not discoverable using ICD-9 codes, which are commonly used in large public databases like the National Inpatient Sample, National Cancer Database, the Centers for Disease Control and Prevention National Program of Cancer Registries, and the Surveillance, Epidemiology and End Results Program at the National Cancer Institute. Based on the findings of this paper, investigations using ICD-9 codes will likely result in underrepresentation of the variables in question, which raises general questions about the validity of these large public databases in studying cancer prevalence, treatment, and outcomes. We also don’t know how generalizable the NLP method will be across institutional databases for sarcoma research or what the next steps should be. For example, once patients with specific diagnoses are identified, how can NLP—or AI, in a broader sense—be used to integrate essential and analyzable data into a coherent and comprehensive database to study these rare diseases? Nonetheless, the work of Lindsay et al. [4] represents an exciting step in an important direction.

Patient-reported outcomes are used to assess parameters like physical function, emotional and psychological well-being, and overall acceptance of a therapeutic intervention. The problem is that patients with orthopaedic oncologic problems experience physical, emotional, and financial stresses that may exceed those observed among others who are treated for more-elective musculoskeletal problems. These patients have many visits with physicians and allied care professionals and receive many invitations to participate in research studies and clinical trials—all of which drains a patient’s time, energy, and patience with surveys. Additionally, many of these survey questions are redundant and confusing. For example, how can one quantify “emotional acceptance of a procedure” on a 1 to 5 scale when a large part of your skeleton has been removed or radiated while receiving chemotherapy that affects your memory, immune system, and vital organs—all while your mortality hangs in the balance? Reducing survey fatigue will increase our collective research success, and for this reason, I congratulate Nalty et al. [6], whose work in this month’s CORR offers guidance on how we might do this. Their short-form survey, which was abstracted from two validated, long-form surveys, demonstrated similar sensitivity and specificity to the more-burdensome original.

I’m excited by what’s possible in the coming years, and I believe that if you dip into this year’s selected MSTS proceedings in CORR, you will be, too.