Background
Osteosarcoma is one of the most malignant human primary bone cancers, which mainly affects children and adolescents with a median age of about 20 years old. The incidence rate is rare in China with only about 3 per million persons every year. This fact is probably the reason why few data of osteosarcoma was documented or analyzed referring to the prognosis in this region.
Advances in the clinical management of osteosarcoma, specifically the introduction of multiagent chemotherapy in combination with surgery, had led to dramatic prognostic improvements in 5-year survival rates from less than 20% to 55–80% in the 1980s [
1,
2]. Several international studies have demonstrated that MAP (high-dose methotrexate, doxorubicin, and cisplatin) regimen was the most effective chemotherapy against osteosarcoma [
3‐
5]. However, additional improvements in survival rates have not been achieved in the last 30 years [
6], and there are still controversies in some aspects concerning prognostic evaluation and therapeutic approaches [
2,
7‐
21]. For example, it remains unknown whether elevated serum alkaline phosphatase (ALP) or lactate dehydrogenase (LDH) can be regarded as independent prognostic factors. In the extremity osteosarcomas, no consensus has been reached on whether patients can benefit from amputation or limb salvage surgery. Furthermore, very limited information is known about survival and prognostic factors in patients across all subgroups of osteosarcomas because most of the previous studies predominantly focused on “classic osteosarcoma”, which should merit the whole criteria of age < 40 years old, high-grade histology, extremity localized primary tumor, and no detectable metastasis at primary diagnosis [
22]. The “non-classic osteosarcoma” was defined as primary metastatic disease, non-extremity localization, or age > 40 years [
22].
Therefore, we conducted a retrospective study covering both classic and non-classic osteosarcoma in our center, which is located in the South China. The aim of this study is to find out the prognostic factors of osteosarcoma and to analyze the overall survival (OS) and event-free survival (EFS) in different patient subgroups.
Discussion
Osteosarcoma is a kind of solid tumor with relatively low incidence, which up to now lacks clinical documentary. Most of the previous studies only referred to the classic osteosarcomas, while few articles have focused on the prognosis of the non-classics. The present study enrolled both classic and non-classic osteosarcoma patients with histologically proven high-grade, and all the enrolled patients were treated with the uniform strategies with neoadjuvant chemotherapy (MAP regimen), surgery, and adjuvant chemotherapy (MAP regimen). The uniform treatment eliminates the bias associated with therapeutic regimen, making comparisons valid.
Our survival results are similar to those in other previous studies. A single-institution study from Malaysia recruiting 163 patients reported 5-year OS of 44% and 5-year DFS (disease-free survival) of 33% [
27]. The Brazilian osteosarcoma treatment group studies III and IV showed that 5-year OS was 50.1% and 5-year EFS was 39% [
13]. An American team using Surveillance, Epidemiology, and End Results (SEER) program database to analyze the survival in 2849 patients with high-grade osteosarcoma presented 5-year OS of 71.8% in non-metastatic disease and 30.4% in metastatic disease at diagnosis [
28]. Collectively, the reviewed studies showed that 5-year OS and EFS ranged from 37.9 to 77% and 38 to 62%, respectively [
2,
7‐
21]. Consistent with these previous studies, the survival data in this study were located within the same range as reported before.
In this study, we enrolled primary tumor site and metastasis status at diagnosis, as well as gender, age, tumor size, symptom duration, type of surgery, and serum ALP and LDH levels for prognostic analysis, some of which have also been analyzed in other previous studies, but some of the results are still controversial. In Table
4, we listed some similar studies as ours.
Table 4
Review of prognostic factors for OS and/or EFS in previous studies by univariate and multivariate analysis
Gender | ns | – | s | s | ns | – | ns | – | ns | – | s | s | s | s | s | ns | – | – |
Age | ns | – | ns | – | s | s | s | s | s | ns | s | s | ns | – | s | s | – | – |
Tumor site | ns | – | s | s | s | s | s | s | s | s | s | s | ns | – | – | – | – | – |
Tumor size | s | ns | – | – | – | – | – | – | ns | – | s | s | – | – | – | – | s | ns |
Symptom duration | – | – | – | – | – | – | – | – | ns | – | – | – | – | – | – | – | – | – |
Primary metastasis | s | s | – | – | s | s | s | s | s | s | s | s | – | – | s | s | s | s |
Serum ALP | s | s | – | – | – | – | – | – | – | – | – | – | s | s | – | – | – | – |
Serum LDH | ns | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – |
Surgery type | – | – | s | ns | s | s | – | – | – | – | – | – | ns | – | – | – | s | – |
Our study showed that primary metastasis status at diagnosis was an independent prognostic factor. Seven of the 9 studies evaluated primary metastasis status, and all validated it as an independent prognostic factor for the outcome of osteosarcoma, which is in fact a foregone conclusion. The study of American SEER program database revealed primary metastasis as an independent predictor for cause-specific survival at 10 years [
28]. Pakos et al. [
15] showed that the death risk of metastasis at diagnosis is nearly three folds than that of non-metastasis, while Janeway et al. [
17] reported that the presence of metastasis at diagnosis had a poorer survival in patients with high-grade osteosarcoma of any site enrolled on North American Cooperative Group. Similarly, the other four studies demonstrated that clinically detectable primary metastasis were significantly associated with inferior outcome [
9,
13,
18,
27].
Primary tumor site was another independent osteosarcoma prognostic factor in our study, which was supported by most of the listed studies in Table
4 [
2,
15,
17,
18,
28]. It is worth noting that Pruksakorn et al. [
18], Duchman et al. [
28], and Min et al. [
8] used the same categorization as ours in the analysis of tumor site which compared extremity tumors with axial tumors. Consistent with our data, Duchman et al. [
28] and Pruksakorn et al. [
18] reported that axial osteosarcomas have lower survival rate than the extremities. With different categorizations, Whelan et al. [
2] showed that patients of distal tumor location had better survival than those of proximal humerus/femur, Pakos et al. [
15] showed that the death risk of tibia tumor location was lower than that of femur tumor location, and Janeway et al. [
17] reported that pelvic tumor site was associated with a decrease in EFS and OS.
Other variables of gender, age, tumor size, duration of symptoms, type of surgery, and serum ALP and LDH levels analyzed in our study did not show predictive values for OS or EFS, and some coincident negative results can also be found in other studies (Table
4). These studies, except that of Petrilli et al. [
13], evaluated gender in univariate analyses, but only four reported significant association with outcome, three of which suggested that female patients had a better outcome than male patients in the multivariate analyses. Hence, whether gender is an independent prognostic factor for patients with osteosarcoma remains controversial. There are eight of these studies evaluated the predictive value of age, and only four studies showed elder age was associated with worse outcome of osteosarcoma. However, it must be noted that they used different cut-off points, i.e., 18 years in the study of Janeway et al. [
17] and 60 years in the study of Duchman et al. [
28]. Converted to ranked data without cut-off points, elder age was reported as an adverse prognostic factor with 7% relative risk increasing per decade by Pakos et al. [
15]. Faisham et al. [
27] showed that patients older than 12 years had worse survival, but no significant difference was found when the cut-off point was 40 years. Referring to tumor size, only Duchman et al. [
28] reported it as an independent risk factor which was divided into small (diameter ≤ 5 cm), intermedia (5–10 cm), large (> 10 cm), or unknown. Other three studies used different categorizations (i.e., diameter of 12 cm [
9,
13] or tumor volume of 180 ml [
18] as cut-off points) but obtained no significant result. Our study categorized tumor size by the diameter of 8 cm according to TNM Staging system [
24], which was the same as the studies by Nathan and Healey [
16] and Wang et al. [
20], and we did not find significant difference between these two groups. Of these nine studies, only Prusakorn et al. [
18] evaluated the symptom duration before diagnosis, and the result showed that it was not a predictor for the outcome of ostnon-classic osteosarcoma, which was the same as our data. There were few researches focusing on the effect of ALP or LDH. Only Vasquez et al. [
9] examined the predictive value of LDH and found that it was not a prognostic factor. Vasquez et al. [
9] and Min et al. [
8] reported that raised ALP was associated with decreased prognosis. Type of surgery was evaluated by four studies, while only the data of Pakos et al. [
15] revealed it as an independent prognostic factor, which showed that the risks of metastasis and death would increase in patients undergoing an amputation. Given the different conclusions reported in literature, we speculated that this might result from patient heterogeneity, different cut-off values, and small sample sizes in most of the studies. Therein, different demographic backgrounds, tumor-related characteristics, and therapy protocols account for the heterogeneity of patients. The subjects of these studies are of different races, gender proportions, and age distributions, which may also lead to varied characters of tumor genesis, progression, and outcome. These at least partly explain why most of the studies are limited to patients with classic osteosarcoma, which is the common form of osteosarcoma with almost uniform treatment.