Prognostic variables in osteosarcoma: a multi-institutional study.

Abstract

This is a report of a multi-institutional study of all patients with osteosarcoma who were seen at 13 comprehensive cancer centers from July 1, 1977, to December 31, 1982. Follow-up extended to 9 years; a minimum of 3 years was obtained for greater than 90% of the patients. All patients with osteosarcoma were considered, but only those with tissue confirmation who had had at least part of their first course of treatment at one of the 13 institutions were included. There were 543 patients. In a search for prognostic indicators, 38 patient characteristics, three treatment categories, and an institutional variable were studied. A combination of nine of these constituted the best indicator of survival. They were morphology (two parts), site of primary cancer (two parts), spread of tumor, grade and size of tumor, duration of symptoms, weight loss of greater than 4.5 kg (10 lb), swelling at primary site, and lytic appearance. Unexpectedly, treatment was not one of the indicators of survival. A prognostic score was developed in which the coefficients were obtained from the Cox regression (step-down) method. Each patient had a score (S) and an observed survival time that together provided the expected risk of death for that patient. Although this was not a randomized study, treatments were compared before and after adjusting for characteristics identified as prognostic. Three treatments differed little: surgery alone, surgery plus chemotherapy and/or radiotherapy, and chemotherapy and/or radiotherapy followed in 1-4 months by surgery. Patients with amputations and those with resections had similar death rates, but the observed progression rates differed widely. However, when the rates were adjusted for prognostic characteristics, the difference disappeared. Complete surgery (if osteosarcoma existed within surgical margins) was no better than incomplete surgery (if osteosarcoma existed beyond surgical margins) with respect to death but, as would be expected, complete surgery was much better with respect to disease progression.