Abstract
Background/Objectives: Tumor-induced osteomalacia (TIO) is a rare acquired paraneoplastic syndrome caused by phosphaturic mesenchymal tumors (PMTs). FGF23, which is overproduced by PMTs, causes hypophosphatemia and osteomalacia, ultimately leading to multiple insufficiency fractures, which are the cause of TIO symptoms. Therefore, recovery from TIO symptoms often takes several months. Due to its paracrine effects, even minuscule amounts of residual PMT can cause treatment to fail. To further compound this, the most confident methods for residual PMTs, serum FGF23 level and (68)Ga DOTA-based PET/CT, are not readily available. For these reasons, there is currently no established method for early prediction of TIO treatment outcomes after surgery. This study focuses on mineral metabolism and bone turnover markers to identify a clinically practical and readily available biomarker that can predict TIO treatment outcomes. Methods: During treatment, we analyzed repeated measurements during treatment of mineral metabolism and bone turnover markers for 19 cases of TIO-Ca, inorganic phosphate (Pi), parathyroid hormone (PTH), 25-hydroxyvitamin D, alkaline phosphatase, Procollagen 1 N-terminal Polypeptide, and β-CrossLaps-in relation to treatment outcomes. We selected predictive marker candidates from among these markers by analyzing their patterns of change during treatment based on three viewpoints-association with (1) cure status, (2) time after treatment, and (3) the interaction effects between (1) and (2) using Linear Mixed Model analysis. We also validated the predictive performance of the selected candidates. Results: In long-term follow-up, only serum Pi and PTH levels were significantly associated with all three metrics mentioned above, suggesting that their patterns of change reflect the clinical course and results of TIO treatment. Pi was the only marker that displayed the same associations during short-term follow-up (two weeks and six weeks after treatment), suggesting that it is a rapidly responsive marker. The serum Pi level two weeks after treatment (Odds Ratio = 7.314, p = 0.028, AUC value of 0.907) and the normalization of Pi at two weeks post-treatment (Relative Risk = 9.975, p = 0.010; sensitivity = 100.0% [95% Confidence Interval (CI) 0.860 to 1.000], specificity = 60.0% [95% CI, 0.208 to 0.600]) were both significantly associated with a complete cure. Conclusions: Serum Pi is a fast, simple, and highly sensitive marker that can replace serum FGF23 and (68)Ga DOTA-based PET/CT in clinical practice for predicting a complete cure of TIO within two weeks of surgery.