This study presents detailed incidence and survival data by age, sex, race, and pathology. Prior SEER studies consisted of reports on either all types of bone cancer combined5
or were limited to those occurring in childhood.21, 22, 28
Our study confirms the commonly-observed bimodal age distribution of osteosarcoma incidence.5, 8, 29, 30, 32
Although a recent National Cancer Database report suggested there was no bimodal age distribution, that report was based on absolute case numbers, not population-based incidence rates.4
Our use of a large, population-based data set largely avoided the possibility of selection bias, which limits the interpretation of data from hospital-based studies.
The present study shows that females of each age stratum less than 15 years had slightly higher incidence rates than males, consistent with many earlier reports,10, 16–18, 22, 29
and in contrast to a single report suggesting that females only have higher rates in the 10–14 age stratum.21
We also observed that the first incidence peak occurs earlier in females than in males. This correlates with an earlier onset of puberty and the adolescent growth spurt. These findings suggest that bone growth and/or hormonal changes during puberty may contribute to osteosarcoma pathogenesis.18, 41
In reports on osteosarcoma that combine patients of all ages, it is likely (but unspecified) that osteosarcoma with Paget’s and osteosarcoma occurring after a previous malignancy are included in the incidence and survival rate estimates. Here, we show distinct trends for these subtypes in the elderly. The age-, gender-, site-, and race-specific patterns we observed for osteosarcoma with Paget’s and osteosarcoma occurring after a previous malignancy were similar to those described in the few other reports focused on these neoplasms.42–45
Several reports have indicated that the majority of osteosarcoma in elderly patients occur with Paget’s disease or consider it a secondary lesion,10–13
while other reports suggest that primary osteosarcoma (occurring as the first cancer) occurs more commonly in older patients.43, 46–49
The interpretation of osteosarcoma incidence after a previous malignancy is challenging, since osteosarcoma is a common treatment-related cancer. We were not able to link the osteosarcoma cases occurring as a second or greater cancer to a specific primary cancer type.
We were able to assess 653 elderly patients with osteosarcoma, to our knowledge the largest series yet reported, and found that osteosarcoma with Paget’s and osteosarcoma occurring after a previous malignancy represent only approximately 34% of cases, as compared with the 56% observed by Huvos.13
Two recent studies in Japan46
also found that the majority of elderly osteosarcoma patients had primary osteosarcoma, and suggest that misdiagnosis may cause an overestimation of osteosarcoma as a secondary lesion.46
From 1973 to 2004, the most significant overall change in osteosarcoma incidence in our study was the decline observed in elderly patients, possibly due to changes in reporting Paget’s and secondary osteosarcoma in this age group.
The epidemiologic patterns of osteosarcoma were different among the three age groups. The age groups chosen are based on the overall osteosarcoma incidence peaks and plateau. Osteosarcoma incidence varied uniquely by race and sex among the age groups. Many earlier studies that relied on data for Whites and Blacks only5, 15, 21, 22, 26, 30
suggested that incidence was greatest in the latter, and our data for all ages combined agrees with this conclusion. However, our comparison among the age groups with additional racial designations, as defined by SEER, found that the incidence is actually greatest in the Other race designation among patients < 25 years, specifically in Asian/Pacific Islanders, in agreement with a recent childhood cancer incidence study,33
while in elderly patients incidence was greatest among Whites. In all age groups, osteosarcoma incidence was higher in males, except in middle age and elderly Black patients where the incidence was slightly higher in females. There was a much higher incidence of distant disease in elderly patients compared with the younger age groups, perhaps in part due to delayed diagnosis or age-specific differences in tumor biology.
The anatomic site distributions were more diverse for the 25–59 and 60–85+ age groups compared with the 0–24 group; however, the lower long bones were always the most common site. Among the elderly, there were considerable differences in anatomic location between males and females for osteosarcoma occurring after a previous malignancy. Although we were not able to determine the previous cancer diagnoses for those subjects, it is theoretically possible that the anatomic site differences could be related to differences in the location of their primary cancer and the resultant radiation field; for example, therapeutic radiation for breast cancer could contribute to the observed osteosarcoma increase in the chest region of elderly females.
Differences in incidence and survival were also noted between pathology subtypes, although these analyses are somewhat limited because the majority of osteosarcomas are reported as histology Not Otherwise Specified. In the future, as NOS reporting continues to decrease, we may achieve more insight into the characteristics of pathologic subtypes. We were able to show that parosteal osteosarcoma was more common in females age 0–24 and 25–59, corroborating the National Cancer Database report (all ages combined).4
In addition, we demonstrated that chrondroblastic and parosteal osteosarcoma were more commonly reported in the adolescent and middle age groups, and parosteal and fibroblastic osteosarcoma were significantly more frequent in the middle age group.
Survival after osteosarcoma diagnosis improved significantly for all three age groups between 1973–83 and 1984–93. Our analyses also showed that survival varies by gender, age, anatomic site, pathology sub-type, and disease stage. Survival rates were highest in females and in the youngest age group, and lowest in elderly patients. The greatest indicators of survival in all age groups were anatomic site and disease stage. In all age groups, the highest survival rate was observed with osteosarcoma of the short bones and localized disease, and the poorest with osteosarcoma of the pelvic region and/or vertebral column and distant disease. Survival in patients with tumors in the upper long bones was particularly poor in elderly patients, but not in the other age groups. Osteosarcoma pathology also affected survival, although it is difficult to evaluate the accuracy of these measures, as many of the pathology subtypes were underrepresented (<10 cases) in each age group.
Prior to the 1980s, the osteosarcoma survival rate in the U.S. was approximately 20%.50, 51
Five-year osteosarcoma survival rates for children and adolescents in Europe also show marked improvement up to the 1980s and little improvement thereafter.24, 53
Clinical trials including chemotherapy, given both before and after definitive surgical resection, began in the early 1980s. These studies resulted in rapid improvement in 5-year survival, to approximately 70%. However, improvements in osteosarcoma survival during the last decade have been limited; clearly, new treatment strategies are needed (recent treatment developments reviewed in Ferrari and Palmerini52
We have confirmed that osteosarcoma incidence is bimodal, more common in males, occurs most frequently in the lower long bones, and that survival rates have leveled off since the mid-1980s. We have also, for the first time, demonstrated the impact of osteosarcoma associated with Paget’s disease or occurring as a second malignancy on the incidence and survival of “pure” osteosarcoma. However, among the three age groups there are many unique epidemiologic features, illustrating that these age groups should be studied separately in order to more completely understand osteosarcoma epidemiology and underlying biology.