With the widespread application of deep learning technology in radiotherapy, increasing attention has been directed toward enhancing the precision and personalization of intelligent radiotherapy for osteosarcoma. This review summarizes the recent advances in the application of deep learning to osteosarcoma image preprocessing, automatic target volume delineation, dose prediction and treatment plan optimization, early efficacy assessment, and follow-up monitoring. Bottlenecks such as data sharing, model generalization, and interpretability are analyzed. This review aims to provide a comprehensive reference for technological iteration and clinical implementation in this field, as well as a scientific basis for the determination of future research directions and the development of standardized guidelines.

Both primary (e.g. osteosarcoma and Ewing sarcoma) and secondary (e.g. metastatic) bone cancers cause severe health problems and can lead to death. In recent years, radiopharmaceuticals have shown significant potential in the treatment of refractory bone diseases due to their unique targeting properties and low toxicity. Radiopharmaceutical therapy for bone cancer utilizes bone-seeking radiopharmaceuticals that target lesions through the high affinity of ethylenediaminetetramethylenephosphonic acid (EDTMP) ligand for bone hydroxyapatite. Traditional therapeutic agents such as ⁸⁹Sr and ¹⁵³Sm-EDTMP, administered via intravenous injection, accumulate in bone metastases and release beta rays to kill tumor cells. These agents can relieve pain and inhibit bone destruction; however, they have limited ability to shrink tumor volume or slow tumor growth (tumor volume reduction rate < 30%). Furthermore, these agents may cause systemic side effects, and existing evidence does not support their effectiveness in prolonging survival. The development of palliative bone pain therapies based on novel radiopharmaceuticals may overcome these limitations and improve the quality of life for patients during their remaining lifespan. Researchers have investigated the combination of various beta- and alpha-emitters with novel radionuclide carrier systems, significantly enhancing treatment precision in bone tumor models. For instance, ¹⁷⁷Lu-EDTMP (beta ray 0.497 MeV, half-life 6.7 days) enabled 66% of patients to completely discontinue analgesics by the twelfth week. ¹⁸⁶Re- and ¹⁸⁸Re-hydroxyethylidene diphosphonate, by targeting hydroxyapatite, achieved a pain relief rate of up to 90% and significantly prolonged median survival. These agents caused only reversible hematological toxicity, with no Grade III/IV bone marrow suppression events, and significantly reduced tumor volume. Additionally, the alpha nuclide ²²³RaCl₂ induces DNA breaks in tumor cells and inhibits osteoblast-osteoclast imbalance. When combined with best supportive care, it can extend overall survival by 3.6 months (14.9 months vs. 11.3 months), reduce the risk of death by 30%, and lead to significant tumor volume reduction. However, bone marrow toxicity management, cost-effectiveness balance, and the design of multimodal combination regimens remain key challenges.

Objective To observe the inhibiting activities of fibronectin (FN) and citrullinated fibronectin (cFN) on disintegrin and metalloproteinase with thrombospondin motifs-4 (ADAMTS-4),and to explore the extracellular regulative mechanisms of ADAMTS-4.Methods FN was incubated with peptidylarginine deaminase type 4 (PADI4).Western blotting analysis was used to verify the citrullination of FN.The binding activity of FN and cFN to ADAMTS-4 were investigated by enzyme-linked immunosorbent assay (ELISA).The proteolytic ability of ADAMTS-4 after binding to FN and cFN were measured with the aggrecanase activity assay kit.One-way ANOVA,LSD-t test and t-test were used for statistical analysis.Results The immunosignal of citrulline was detected in FN after incubated with PADI4,but not in the absence of PADI4.A higher absorbance at 405 nm was detected when the full-length ADAMTS-4 protein was incubated with FN (2.182±0.042) than cFN (0.624±0.033; t=50.522,P＜0.01).Additionally,the recombinant ADAMTS-4 protein with a truncation at the C-terminus displayed low absorbance at 405 nm when the enzyme was incubated with both FN(0.971±0.024) and cFN(0.934±0.012; t=2.388,P＞0.05).Large amounts of ARGxx peptide were detected with full-length ADAMTS-4 in aggrecanase activity assay [(0.908±0.088) nmol/L],but significantly less when in the presence of FN and ADAMTS-4 [(0.573±0.000) nmol/L,P＜0.05].The production of this peptide was more when full-length ADAMTS-4 was incubated with cFN [(0.830±0.020) nmol/L,P＜0.05] than with FN.The reaction containing the truncated ADAMTS-4 without FN or cFN yielded the highest concentration of ARGxx peptide [(36.420±3.673) nmol/L],peptide production was not significantly altered when FN [(41.099±0.101) nmol/L] or eFN [(41.064±0.083) nmol/L] were added to the reaction.Conclusion FN could bind to ADAMTS-4 and inhibit its proteolytic activity.After citrullinated by PADI4,the binding activity of cFN is weakened and less inhibition to ADAMTS-4.