Treatment of Osteosarcoma from the spine: A population-based database study

Purpose The treatment of osteosarcoma of the spine remains controversial. Our aim is to explore the treatment of patients with spinal osteosarcoma. Methods We analyzed the date collected 727 spinal osteosarcoma patients from the Surveillance Epidemiology and End Results (SEER) databases between 1973 and 2015. X-tile software was performed to nd the optimal cut-off values of age and economic income. Univariate and Multivariate Cox analyses were used to identify the independent prognostic factors. Logistic regression model was conducted to clear the factors associated to surgical compliance; Kaplan-Meier estimator method was adopted to analyze the Overall survival (OS) and Cancer-specic survival (CSS). patients group


Introduction
Osteosarcoma is the most common primary malignant bone tumor (1). The characterized of osteosarcoma is the proliferation of tumor cells that directly form immature bone or bone-like tissue (2). Currently, the global annual incidence of osteosarcoma is approximately 1 to 3 per 1,000,000 population (3,4). Osteosarcomas tend to occur in adolescents and has been the second leading cause of cancer-related deaths in pediatric age group (5). Osteosarcoma affects more men than women (6). The most common site of osteosarcoma is the metaphysis of long bones and less often the skull, jaw, or pelvis (7). Osteosarcoma that occurs in the spine has a poor prognosis compared to the high incidence of osteosarcoma of the limbs, accounting for only 3% of all primary osteosarcomas (8). Since osteosarcoma has high malignant degree, poor prognosis, high mortality and easy to occur distant metastasis, which has a deep impact on society (9).
Surgical treatment is one of the rst choices for the treatment of osteosarcoma. Amputation has been the gold standard for treating osteosarcoma (10). However, the overall 5-year survival rate is low, and most patients die within 1 year after diagnosis (11). In recent years, with the advent of neoadjuvant chemotherapy, the safety of limb salvage surgery has been improved. Limb salvage surgery is generally used in combination with chemotherapy, which has obvious curative effects on tumors, and plays a very important role in clinical practice. It has gradually replaced amputation as the rst choice and mainstream surgery for limb osteosarcoma. 80 to 90 of the 100 patients underwent limb salvage surgery, and the disease-free survival probability also increased from about 20-60% (12,13).However, whole tumor resection is not easy to use for the treatment of spinal osteosarcoma, and spinal osteosarcoma has strong local invasiveness and high local recurrence rate. Moreover, spinal osteosarcoma may spread through early blood-borne metastases, and the effect of surgical resection on prolonging survival is unclear (14).
In previous case series, factors such as age at diagnosis, tumor size and location, pathology, presence and location of metastases, surgical strategy, surgical margins and histological responses, and acceptance of chemotherapy were reported to affect overall survival (15)(16)(17). The effect of surgical treatment of spinal osteosarcoma on overall survival is currently unclear, due to the small proportion of patients with spinal osteosarcoma. The treatment of spinal osteosarcoma still confuses many surgeons, even experienced surgeons. In our study, we used data from the SEER cancer registry to explore treatment options for patients with spinal osteosarcoma to improve the prognosis of osteosarcoma.

Data source and patients
The data presented in this paper were retrieved from the Surveillance Epidemiology and End Results (SEER) database, funded by the National Cancer Institute. The current SEER database includes 18 population-based cancer registries acquired between 1973 and 2015, which represent patient demographics and cancer characteristics for about 28 percent of the U.S. population (18). SEER data can be published for cancer-based epidemiological studies and survival analysis. All case data was retrieved using the SEER*Stat application (version 8.3.5).

Statistical analysis
Basic information of all selected patients, including diagnosis time, gender, race, age, marital status, and disease-related information, such as radiotherapy, chemotherapy, tumor staging, pathological grade, and treatment methods, can be extracted from the database. According to the chronological order of diagnosis, the patients in the SEER database with a diagnosis span of 42 years were divided into 4 groups. Patients were divided into two groups according to different treatment schemes. We refer to patients undergoing surgery as the surgical treatment group, and those who have not undergone surgery are the non-surgical treatment group.
Statistical analysis was performed using Statistical Program for Social Sciences (SPSS) software version 24. Chi-square analysis was used to evaluate the demographic and clinical characteristics of patients and their correlation with treatment. Kaplan-Meier curve was used to estimate the survival time of each group, and the difference between the curves was analyzed by log-rank test. The Overall survival (OS) time is from the date of diagnosis to death from any reason or the date on which data were censored. Cancer-speci c survival (CSS) is a net survival indicator that estimates the likelihood of osteosarcoma survival of the spine in our study. Univariate and multivariate Cox proportional regression models were performed to estimate the hazard ratios (HR) and 95% con dence intervals (CI) to analyze independent prognostic factors. X-tile software was conducted to nd the optimal hierarchical age at diagnosis. Pvalues of less than 0.05 were considered statistically signi cant.

Identi cation of cutoff values for age
To determine the optimal age strati cation for patients with osteosarcoma of the spine, we constructed an X-tile plot to explore the predicted cutoff values. The age of diagnosis was divided into 3 levels: <48 years, 49-62 years, and> 62 years ( Figure 2).

Baseline characteristics of patients
Using chi-square test, we studied the demographic and clinicopathological characteristics of patients with osteosarcoma of the spine. The baseline characteristics of the patients are reviewed in Table I. After applying exclusion criteria, a total of 727 patients were included in our cohort from 1973 to 2015. Among them, 370 cases were treated by surgery and 357 cases were not treated.
Chi-square test showed that different treatment patterns was related to age at diagnosis (p<0.001), marital status (p<0.001), SEER historic stage (p<0.001), tumor grade (p<0.001) radiotherapy (p<0.001) and chemotherapy (p<0.001). Patients younger than 48 years of age choose surgical treatment, while patients older than 62 years old prefer non-surgical treatment. In terms of marital status, married patients accounted for a relatively high proportion (48.0%) of the patients who chose surgery, and divorced or separated patients had poor surgical compliance (4.0%). In SEER tumor staging, patients with regional osteosarcoma of the spine were more likely to receive surgical treatment than patients with distant. As to the chemotherapy, the proportion of patients who received chemotherapy chose surgery treatment was signi cantly higher than that of patients with no chemotherapy. However, this situation is exactly the opposite in radiotherapy.

Survival analysis
Comparison of the survival outcome between different treatment groups By analyzing the Kaplan-Meier curve with a log-rank test, we found that age at diagnosis (p<0.001), Year of diagnosis (p=0.008) marital status (p=0.002), SEER historic stage (p<0.001), tumor grade (p<0.001), surgery (p<0.001), radiotherapy (p<0.001) and chemotherapy (p<0.001) were associated with OS (Table  II). However, marital status and chemotherapy treatment were not associated with CSS (Table II). The OS median survival time of patients undergoing surgery was 27 months and that of patients without surgery was 8 months (Table II and Figure 3a). For CSS median survival time, patients undergoing surgery was 44 months and patients without surgery was 11 months (Table II and Figure 3b). In both OS and CSS, surgically treated patients had signi cantly longer survival time than non-surgically treated patients ( Figure 3).
Objective to explore which variables have signi cant in uence on the prognosis of patients with osteosarcoma of the spine. We further used multivariate analysis to con rm that surgery was an independent prognostic factor for OS and CSS. Compared to patients with surgery treatment, the risk was increased when the patients without surgery in the OS (HR= 1.66, 95%CI: 1.36-2.02, P < 0.001) and CSS (HR= 1.56, 95%CI: 1.24-1.97, P < 0.001). Moreover, for age and SEER historic stage, multivariate analysis also showed statistically signi cant difference in both OS and CSS.
Factors willing to undergo surgery We used multivariate logistic regression model to determine the factors that in uence whether to undergo surgery. When adjusting for other factors, some variables were proved to be signi cantly correlated with rejection of surgery ( Figure 4). We found that the in uencing factors have the following aspects: age, grade, and SEER historic stage. Patients who were 62 years or older (OR, 3.86; 95% CI,2.63-5.56; P < 0.001), at regional stage (OR, 6.05; 95% CI,3.71-9.87; P < 0.001) and at unstaged stage (OR,4.62; 95% CI,2.45-8.72; P < 0.001). Additionally, Patients who were at histological grade (OR, 0.79; 95% CI,0.14-4.62; P =0.797) were more willing to accept surgical therapy.
Trends in survival strati ed by the age at diagnosis and SEER historic stage In order to better demonstrate that surgery as a single factor had a signi cant impact on the survival of patients with osteosarcoma of the spine. Patients were strati ed by the age at diagnosis and SEER historic stage to investigate the trends in survival. For OS and CSS, we found that regardless of the age stage, patients in surgery group had more favorable survival than the other group ( Figure 5). Similarly, the outcome in SEER historic stage had the same characteristics ( Figure 6).

Discussion
Osteosarcoma of the spine is a rare malignant tumor with the potential for local invasive destruction and systemic metastasis, and is considered to have a poor prognosis (19,20). Osteosarcoma mainly occurs in adolescents or young adults, which seriously damages social productivity (21). Surgery is one of the effective treatments for osteosarcoma (22), but osteosarcoma of the spine is excluded in a considerable proportion of cases due to its aggressiveness and the special anatomical structure adjacent to the spinal cord and nerve roots (23). Currently, there are few studies on the optimal treatment for osteosarcoma of the spine. Whether surgical treatment is appropriate for osteosarcoma of the spine remains controversial. Therefore, it is of great signi cance to conduct the studies.
Surgery is considered to be an effective way to treat some malignant tumors, which is derived from some characteristics of the tumor itself. Many literatures have con rmed the rationality of this treatment. Khan reports that primary tumors could lead to metastatic spread, and surgery could reduce the burden on the tumor, thereby reducing the possibility of metastatic spread(24). Cook AD et al believed that reducing tumor load may reduce the occurrence of serious complications such as hypoproteinemia and cachexia, thus reducing the risk of cancer death(25). In addition, Surgical removal of the tumor results in a reduction in volume, which can improve the effect of later chemotherapy and reduce the risk of local recurrence(26). These theories con rm the mechanism by which surgery can improve tumor patient survival.
Similarly, we considered that surgery can improve the survival time of patients with osteosarcoma of the spine. The outcomes of study were in line with our conjecture. According to our criteria, a total of 727 patients were included, of which 370 (50.9%) patients had surgery and 357 (49.1%) patients had nonsurgical treatment. Survival analysis con rmed that surgical treatment has a signi cant impact on patient prognosis. In our study, survival of patients undergoing surgery is signi cantly higher than that of patients not undergoing surgery. Many clinical studies support aggressive surgical treatment of patients with osteosarcoma of the spine. Surgical patients have a better prognosis, which is consistent with our ndings. Ozaki et al evaluated 22 patients with osteosarcoma of the spine and found that the survival of the 5 patients who underwent surgical resection was signi cantly different from the 17 patients without resection(27). Debraj Mukherjee et al studied 158 patients with osteosarcoma of the spine and found that patients with surgical resection had a better survival advantage(28).
Although studies have shown that age increase is independently related to poor survival(29), in our study it was con rmed that regardless of age, the survival time of surgically treated patients is always higher than that of non-surgical resection. In the same way, even if the stage of osteosarcoma of the spine affects the patient's prognosis(30), our research shows that the effect of surgery on patient survival is bene cial and is an important factor affecting the prognosis. Combining multiple demographic and clinicopathological factors, surgery is still an independent factor. The ndings of Pan Y et al also support our point(31). Surgical treatment has a positive impact on the prognosis of patients with osteosarcoma of the spine.
We used logistic regression analysis to nd out the effect of age, grade, and SEER historic stage on patients' choice of surgery. Our study found that patients with lower age groups, lower tumor grades, and localized tumor distribution were more likely to undergo surgery. In addition, multiple groups of studies found that age tumor grade and distribution were independent factors affecting prognosis(31, 32). The patients with lower age group, lower tumor grade, and localized tumor distribution have better prognosis. We believe this is due to the fact that these patients received surgical treatment, which resulted in longer survival.
This study has several limitations. First, this study is retrospective and has obvious limitations. Second, the SEER database lacks speci c information about the operation, such as the surgical method and the extent of surgical resection. Moreover, because the database cannot accurately provide speci c treatment methods, the effects of radiation and chemotherapy on tumors cannot be considered comprehensively. Despite this, the study is based on a fairly large demographic and is still very convincing.

Conclusions
In our research, the SEER database was used to assess the impact of surgery on survival in patients with osteosarcoma of the spine. Surgery can provide survival bene ts for patients with osteosarcoma of the spine. Spinal osteosarcoma patients with undergone surgery have favorable survival and surgery can become a suitable treatment for patients.
Declarations 24. Khan SA. Surgery for the intact primary and stage IV breast cancer… lacking "robust evidence". The cutoff values of age and economic income were determined by X-tile program.
Percentages may not total 100 because of rounding. Percentages may not total 100 because of rounding. The cutoff values of age and economic income were determined by X-tile program.
After univariate analysis, we selected variables with P < 0.1 for further multivariate analysis. At the same time, we will also consider the impact of clinical practice. The cutoff values of age and economic income were determined by X-tile program.
After univariate analysis, we selected variables with P < 0.1 for further multivariate analysis. At the same time, we will also consider the impact of clinical practice. Figure 1 Schematic ow diagram of inclusion and exclusion criteria for our study cohort.

Figure 2
The X-tile analysis was used to identify the optimal cutoff values of age of diagnosis.

Figure 3
Kaplan-Meier estimates of the Overall survival (a) and Cancer-speci c survival (b) for the total cohort among two groups (surgery group; non-surgical group).

Figure 4
Forest plot of Multivariable Logistic analyses of surgical noncompliance adjusted by age, grade and SEER historic stage. The green squares on the transverse lines represent the Odds ratio (OR), and the transverse lines represent 95% CI. The cut-off values of age and economic income were determined by Xtile program. Kaplan-Meier estimates of the Overall survival and Cancer-speci c survival for the patients diagnosed in different age among two groups (surgery group; non-surgical group). The cut-off values of age and economic income were determined by X-tile program. In the left column, a, c, e is the Overall survival of patients diagnosed at different ages in the two groups (surgical group; non-surgical group). In the right column, b, d, f is the Cancer-speci c survival of patients diagnosed at different ages in the two groups (surgical group; non-surgical group).

Figure 6
Kaplan-Meier estimates of the Overall survival and Cancer-speci c survival for the patients diagnosed in different SEER historic stage among two groups (surgery group; non-surgical group). The cut-off values of age and economic income were determined by X-tile program. In the left column, a, c, e, j is the Overall survival of patients diagnosed at different SEER historic stage in the two groups (surgical group; nonsurgical group). In the right column, b, d, f, h is the Cancer-speci c survival of patients diagnosed at different SEER historic stage in the two groups (surgical group; non-surgical group).