Can We Develop a Web-based Nomogram That Could Predict the Risk of Local Recurrence After Treatment for Giant Cell Tumor of Bone of the Extremities by Curettage?

Background: For giant cell tumor of bone (GCTB) in the extremities, joint-preserving surgery (curettage) is often favored over en bloc resection to maintain postoperative function, although it carries a notable risk of local recurrence. We decided to develop a nomogram to predict the risk of local recurrence after curettage of GCTB of the extremities so physicians can shorten the interval between radiographs, add CT or MRI, and even monitor tartrate-resistant acid phosphatase 5b levels during surveillance for patients at particularly high risk of local recurrence. Questions/purposes: Can we develop a web-based nomogram that could predict the risk of local recurrence after treatment for GCTB of the extremities by curettage? Methods: A total of 535 patients were histologically diagnosed with GCTB of the extremities and underwent surgery at the authors' institutions between 1980 and 2022. Of these, 38% (204) underwent en bloc resection, five underwent amputation, and one patient with an unknown Campanacci stage was excluded. A retrospective analysis was performed on the remaining 325 patients who underwent curettage. The median age was 30 years, and 67% (216) of the patients had a tumor of the distal femur or proximal tibia. The incidence of Campanacci Stage III tumors was 21%, and pathologic fractures were observed in 6% (19) of the patients at presentation. Nomograms were developed using the multivariate Cox regression model to allow prediction of local recurrence rates at 2 and 5 years; covariates in the Cox regression included prognostic factors considered clinically significant, such as patient age, Campanacci stage, site, presence of pathologic fracture at presentation, preoperative denosumab therapy, previous surgery, filler (whether or not bone cement was used), and local adjuvant therapy, and these were considered potential prognostic factors in the model. Model validation was performed using calibration (calibration plots) and discrimination (Harrell C-index). Results: The local recurrence-free survival rate was 83% (95% confidence interval [CI] 79% to 87%) at 2 years and 77% (95% CI 72% to 82%) at 5 years. Calibration plots for local recurrence rates at 2 years and 5 years showed good linearity and general agreement between observed and predicted local recurrence rates. The optimism-corrected C-index of the predictive model was 0.69 (95% CI 0.65 to 0.78). We therefore developed a web-based application incorporating this nomogram, which we then made freely available as a prognostic tool for clinicians, accessible on our webpage: https://naraseikei.com/research/nomogram/. Conclusion: The nomograms we developed are reliable methods useful for predicting local recurrence in patients after curettage of GCTB of the extremities. By using this nomogram, patients at high risk of local recurrence after curettage may be identified, and efforts to detect local recurrence early and perform recurettage or radiofrequency ablation may improve the frequency of joint preservation. Furthermore, the risk of local recurrence can be determined as a specific numeric value before treatment begins, both with and without preoperative denosumab therapy, bone cement use, and local adjuvant therapy, allowing for discussions with patients regarding treatment plans, including the advantages and disadvantages of each option. We note that we did not study whether or not early detection of local recurrence or treating patients with a predicted higher risk of local recurrence will lead to a lesser likelihood of local recurrence or better joint preservation. External validation using data from other institutes is needed. Level of evidence: Level III, prognostic study.