Single fraction stereotactic radiosurgery and fractionated stereotactic radiotherapy provide equal prognosis with overall survival in patients with brain metastases at diagnosis without surgery at primary site

Data source

This study utilized data from the National Cancer Database (NCDB). The NCDB is a venture of both the American College of Surgeons and the American Cancer society. The database is provided with information from hospital registries from Commission on Cancer accredited facilities. More than 70% of newly diagnosed cases of malignant cancer are accounted in the NCDB. Since the data extracted for the study was previously de-identified, there was not a need for Institutional Review Board (IRB) approval.

Study population

The population for this study was comprised of individuals who were 18 years of age or older who had BMs at the time of diagnosis of breast cancer, non-small cell lung cancer (NSCLC), small cell lung cancer (SCLC), other lung cancers, melanoma, colorectal cancer (CRC), or kidney cancer and had received either SRS (dose of 1,500–2,400 Gy (1 fraction)) or SRT (dose of 2,100–3,000 Gy (three fractions), and dose of 2,500–3,250 Gy (5 fractions)) as treatment for their BMs. Patients with diagnoses between 2010 and 2015 in the NCDB were considered for the study. The year 2010 is the year when the NCDB first started collecting information about BMs at the time of the diagnosis of primary cancer. Individuals with surgery at the primary cancer site were excluded from the study because our previous study has shown that these patients belong to a different cohort with a much higher overall survival (Amin et al., 2020a). Information about brain surgery is not reported in the database.

Biological Effective Dose (BED) calculation

BED₁₀ of 24 Gy in 1 fraction is 81.6 Gy

BED₁₀ of 18 Gy in 1 fraction is 50.4 Gy

BED₁₀ of 15 Gy in 1 fraction is 37.5 Gy

BED₁₀ of 27 Gy in 3 fraction is 51.3 Gy

BED₁₀ of 30 Gy in 5 fraction is 48 Gy

End points

Overall survival (OS) was the primary outcome for the study. OS was calculated from the date of diagnosis until the date of the individual’s death. Those alive or lost to follow-up were censored. Predictors of receiving SRS rather than SRT were also included in the study. These predictors were reported using the odds ratio (OR).

Predictors or explanatory variables

OS was evaluated and analyzed between the groups receiving SRT as compared to SRS. Other variables of interest were age at diagnosis were age at diagnosis, sex, race, education level, income, place of living, hospital type, insurance status, Charlson/Deyo score, chemotherapy treatment, the type of primary cancer, and the year of the initial diagnosis. The predictors for whether SRS or SRT treatment was received include sex, race, education, income, place of living, hospital type, insurance status, Charlson/Deyo score, chemotherapy treatment, immunotherapy treatment, primary cancer type, and the year of primary cancer diagnosis.

Statistical analyses

Descriptive statistics for categorical and continuous variables are reported. Predictors of SRS use as compared to SRT were identified using the logistic regression model. The OR was reported as the measure of association with likelihood of using SRS. Survival time was measured in months from the date of diagnosis to the date of death. We used the Kaplan–Meier (KM) method to generate survival curves and analyzed the differences between groups using the log-rank test. Cox proportional hazards regression analysis was conducted to estimate the hazard ratio (HR). Univariable Cox proportional hazard ratios were first obtained, and multivariable Cox proportional hazard ratios were determined including variables that had a significance of P < .15 in the univariable analysis. Analyses were conducted using SAS 9.4 (SAS Institute Inc.).

Patient’s characteristics

The study’s population had a total of 6,961 participants with 3,519 (50.6%) being male. All participants fell between the ages of 21 and 90 with a median age of 65.0 years old. Among all patients, 5,423 (77.9%) received SRS while 1,538 (22.1%) received SRT. The median (range) of ages in patients receiving SRS treatment was 65 (21–90) years old, and the median (range) of ages in patients receiving SRT was 65 (24–90) years old. Of the 6,961 participants, 5,887 (85.4%) were white, 743 (10.8%) were black, and 267 (3.9%) belonged to other races. The majority of participants had an income of greater than $35,000 per year (61.3%), lived in an urban setting (98.3%), had health insurance (97.0%), had a Charlson/Deyo score of 0 (67.0%), and had received chemotherapy (72.2%) but not immunotherapy (92.4%). Of the primary cancer types, 171 (2.5%) had breast cancer, 5,595 (80.4%) had NSCLC, 204 (2.9%) has SCLC, 222 (3.2%) has other types of lung cancer, 449 (6.5%) had melanoma, 67 (1.0%) had CRC, and 253 (3.6%) had kidney cancer. All of the study’s demographic data is presented in Table 1.

Factors associated with the use of SRS vs. SRT

Using univariable logistic regression analysis to search for factors found to have an association with the type of treatment received (SRS or SRT) yielded several significant results (Table 2). Individuals with either a Charlson/Deyo Score of 1 or 2 or more were significantly more likely to receive SRS treatment than those who have a score of 0. Patients with certain types of primary cancers (Other lung, melanoma, and CRC) were significantly more likely to receive SRT treatment when compared to kidney cancer. Other factors significantly linked with receiving SRS treatment as opposed to SRT treatment were receiving chemotherapy treatment and having diagnosis before 2014. Age, sex, race, education level, income, place of living, hospital type, insurance status, other types of primary cancers. (NSCLC and SCLC) were not found to have a significant association with SRS or SRT received.

Multivariable logistic regression analysis was conducted while controlling for the significant factors from the univariable logistic regression analysis. Individuals with a Charlson/Deyo score of 1 or 2 or more, a certain primary cancer (Other lung, melanoma, and CRC) were both significantly linked with receiving SRT treatment when compared to kidney cancer. Having a diagnosis before 2014 was significantly linked to receiving SRS treatment. Education level of the area the individual was from, income, place of living, and other primary cancers (NSCLC and SCLC) were not found to have an association with the use of SRS when compared to the use of SRT.

Survival analysis

Comparing the OS of both the individuals receiving SRS and SRT treatments were analyzed using KM curves (Fig. 1). Individuals receiving SRS treatment had a median survival time of 10.9 (95% CI [10.5–11.3]) months. Meanwhile, individuals receiving SRT treatment had a median survival time of 11.3 (95% CI [10.4–12.3]) months. This difference was not found to be significant (Log rank P = 0.31).

Cox proportional hazard analysis

When analyzing the HR using univariable Cox proportional hazard analysis (Table 3), it was determined that receiving SRS treatment was not associated to significantly different survival outcomes (HR: 0.968, 95% CI [0.908–1.032]; P = .31) when compared to receiving SRT treatment. A younger age at diagnosis, being female and an individual belonging to a race that was neither white nor black had a significant association with improved OS. Additional factors with significantly positive association with OS were receiving care from an academic hospital, living in an area with income >$35,000, living in an area with < 13% of people with no high school degree, a lower Charlson/Deyo comorbidity score, prior reception of chemotherapy treatment, certain primary cancers (Breast, NSCLC, SCLC, and melanoma), and a diagnosis in the years 2014–2015 compared to 2010–2013. Meanwhile differences based on race (identifying as black or white), place of living, insurance status, and having other lung cancers (not NSCLC or SCLC) were not found to be associated with OS.

In the multivariable Cox proportional hazard analysis adjusted for factors that were significant in the univariable Cox regression analysis (Table 3), there was no significant difference in OS for individuals who received SRS treatment compared to SRT (HR: 0.942, CI 95% [0.882–1.006]; P = .08). However, female sex, non-white race, younger age at diagnosis, high income-level, receiving treatment at an academic center, and having a primary cancer that is either breast cancer, NSCLC or melanoma, a Charlson/Deyo comorbidity score of 0, receiving chemotherapy, and having a diagnosis date after 2014 were all associated with improved OS.

Further subset-analyses were conducted to determine the associations of SRT vs. SRS with OS in patients with different primary cancer sites. These results (Table 4) showed a significant improved OS for patients with CRC receiving SRT treatment compared to SRS (HR: 0.417, CI 95% [0.204–0.854]; P = .01).

Biological Effective Dose (BED) is an important factor to consider when comparing survival outcomes. For this reason, the association of SRS vs SRT with OS was also analyzed after stratifying SRS and SRT by BED₁₀ of 55 Gy. Additionally, this analysis was further conducted in each primary cancer site (Tables 5 and 6). Univariable and further multivariable Cox proportional analysis revealed no significant differences in OS when comparing patients receiving either SRS or SRT and either above or below a BED₁₀ of 55.