Beyond 5-year survival. A report from the Cooperative Osteosarcoma Study Group (COSS)

Prognostic factors have been well described for osteosarcoma, but analyses evaluating the further course of long-term survivors are lacking. We used the large database of the Cooperative Osteosarcoma Study Group (COSS) to perform such an analysis.

The COSS database 1980-04/2019 was searched for 5-year survivors of primary high-grade central osteosarcoma of the extremities or trunk. Identified patients were analyzed for their further survival outcomes, assessing potentially prognostic and predictive factors already evident at initial disease presentation and treatment as well as their disease course during the first 5 years of follow-up.

Two thousand and nine former eligible patients were identified (median age at initial diagnosis 15.1 (2.5-63.0) years; male vs. female 1149 (57.2%) vs. 860 (42.8%); extremities vs. trunk 1927 (95.9%) vs. 82 (4.1%); extremity primaries <1/3 vs. ≥1/3 of the involved bone 997 (67.8%) vs. 474 (32.2%) (456 unknown); localized vs. primary metastatic 1881 (93.6%) vs. 128 (6.4%); osteosarcoma as a secondary malignancy 41/2009 (2.0%)). Therapy starting by chemotherapy versus primary surgery 1860 (92.6%) versus 149 (7.4%); definitive tumor surgery by limb salvage versus ablative 1347 (67.0%) versus 659 (1 no surgery, 2 unknown); tumor response to preoperative chemotherapy documented for 1765 (94.9%) patients receiving neoadjuvant chemotherapy, good (<10% viable tumor) versus poor 1130 (64.0%) versus 635 (36.0%), local radiotherapy documented for 19 (0.9%) tumors. Recurrence during preceding 5 years no versus yes 1681 (83.7%) versus 328 (16.3%). Median follow-up starting 5 years after initial diagnosis 6.1 (0.002-32.2) years; 1815 survivors and 194 deaths. Overall survival after another 5/10/15/20 years 91.7%/88.9%/85.8%/83.4% for all patients; 97.5%/95.2%/92.4%/89.9% if in remission years 1-5 versus 62.7%/57.3%/53.0%/51.2% if recurrence year 1-5 (p < 0.001). Significant predictors of survival for all patients age at diagnosis (p = 0.038), tumor site (p = 0.030), having experienced the osteosarcoma as secondary malignancy (p < 0.001), tumor response to preoperative chemotherapy (p = 0.002). Multivariate Cox regression testing possible for 1759 (87.6%) patients with complete dataset: Having had a recurrence in years 1-5 (p < 0.001), older age at diagnosis (p = 0.009), and osteosarcoma as secondary malignancy (p = 0.013) retained significance.

Highly important predictors of death such as the extent of tumor response to chemotherapy no longer remain valid after 5-year survival. The individual history of malignancies and their outcomes seems to gain pivotal importance.

This benchmark analysis clearly defined risk factors for the further course of 5-year survivors from osteosarcoma. It argues for large disease-oriented databases as well as for very long follow-up periods. Novel findings will most likely require innovative statistical models to analyze such cohorts.