Does Giant Cell Glioblastoma Have a Better 2 Prognosis in Comparison to Glioblastoma 3 Multiform? A Secondary Analysis of the SEER 4 Database from 1985-2014

: 19 Brain cancer is the tenth leading cause of death in the U.S. Glioblastoma multiforme (GBM) is the 20 most lethal primary malignant central nervous system tumor in adults. The present study employed 21 samples from 1985-2014 to discover the difference in prognosis among glioblastoma subtypes after 22 the evolution of treatment modalities over the past few years. The current study aims to find the 23 differences between Glioblastoma multiforme (GBM) and giant cell glioblastoma (GCG) in terms of 24 prognosis among adults and elderly patients in the U.S. 25 This study is a historical cohort type of study and is conducted on adults and elderly individuals with 26 GBM or GCG from the years 1985-2014 in the U.S. Data were collected from the Surveillance, 27 Epidemiology, and End Results Program (SEER) database. The study exposure was GBM or GCG 28 and the outcome was mortality. The potential confounders were age, sex, race, ethnicity, year of 29 diagnosis, primary site, and surgery. A chi-square test was used for categorical data. A univariate 30 analysis was used for variables having a p-value < 0.05. Potential confounders were selected and 31 evaluated using multivariate logistic regression models to calculate the odds ratio with stepwise 32 selection. 33 The study sample was 25,117. The incidences of GBM and GCG were not similar in relation to age 34 group. Also, Spanish-Hispanic ethnicity was independently protective of GBM and GCG as 35 compared to Non-Spanish-Hispanic ethnicity patients with GBM have a higher mortality rate than 36 do GCG patients. The mortality rate was higher among patients diagnosed before 2010. 37 In conclusion, GCG was not statistically significant in association to reduced mortality. Non-Spanish- 38 Hispanics with GBM or GCG had a higher mortality rate than did Spanish-Hispanics. Factors such 39 as being female, being age >59, and having a year of diagnosis before 2010 were independently 40 associated with increased mortality. 41

diverse population subgroups. These surveys have multi-stage sampling and are considered 77 to be complex, overestimated, and not representative of the entire U.S. population. However, 78 SEER does its own modeling through extrapolation. 79 80

Study population: 81
Patients aged younger than 20 years have a lower incidence rate; frequency 82 rapidly increases starting in the fifth decade of life [13]. Therefore, the inclusion criteria for 83 the analysis were patients with a confirmed diagnosis of GBM or GCG at age 18 or older 84 from the years 1985-2014. The exclusion criteria included insurance, grading, and tumor size, 85 due to a high percentage (over 25%) of missing data in the SEER database. The SEER 86 database included patients' insurance data from the years 2007 and onwards. Also, in terms 87 of tumor size, 65% of data was missing in the database. However, glioblastoma has no clear 88 grading system, as it is a type of glioma and is considered the most malignant type (type 4). 89 Therefore, grading was also excluded [14]. 90

Ethical Considerations 91
Ethical approval was waived, since the analysis was considered nonhuman 92 subjects research by the Florida International University Health Science Institutional 93 Review Board. 94

Study variables: 95
The study variables included data of GBM patients (histology codes: ICD-O-3:9440/3, 96 9441/3) with tumors located in several locations: supratentorial (cerebrum, frontal lobe, 97 temporal lobe, parietal lobe, occipital lobe), brain overlap, and infratentorial (cerebellum, 98 the SEER database. Diagram 1 shows the variables that were analyzed. 100 In addition, the SEER research data record description was used to categorize other variables 101 such as race, which was categorized into White, Black, and Others. Ethnicity was also 102 categorized into Non-Spanish Hispanic-Latino and Spanish-Hispanic-Latino.

Statistical analysis: 106
First, the population was selected from the SEER database. Then, the characteristics of the 107 population were described. After that, the general distribution of the data was examined. 108 Next, some variables were transformed into appropriate categories (e.g. age group was 109 categorized into adults from 18-59 years old and elderly individuals >59 years old) [16]. The 110 primary site was categorized into supratentorial, brain overlap (including the brain ventricles 111 and other unspecified brain locations), and infratentorial regions. 112 The alpha level was set at 0.2 due to the small sample size of GCG incidences in the SEER 113

154
Race also reveals some variations in terms of the two subtypes of glioblastoma, with 155 individuals who have a white racial background being more prone to GBM, while individuals 156 of other races being more prone to GCG. The Non-Spanish-Hispanic-Latino ethnicity has a the incidence of GCG before 2010; after 2010, the incidence of GCG was higher. However, 160 incidences of both tumors have decreased considerably since 2010. 161 The study reveals some statistically significant differences in terms of tumor primary 162 site, with high statistical significance. Both subtypes of tumors originate more often in the 163 supratentorial part of the brain than elsewhere in the central nervous system. However, GCG 164 tumors originate more from the supratentorial site than do GBM tumors. It is also statistically 165 significant that GBM risk is higher in patients with no surgery or no gross total resection, 166 while patients with gross total resection (GTR) have an elevated GCG risk. Table 2 shows 167 that patients with GBM have a higher mortality rate than do GCG patients. Table 3 shows 168 that GCG has an odds ratio [OR] of 0.56 with a confidence interval of 0.53-1.44, which is 169 independently associated with reduced mortality. 170 Table 2 also shows a slight difference in mortality between age groups in relation to the 171 two glioblastoma subtypes; this difference is statistically significant. It indicates that elderly 172 patients have a worse prognosis than do adults. Glioblastoma patients with a white racial 173 background also face a slightly increased risk of death. The Spanish-Hispanic-Latino 174 ethnicity has a lower mortality rate than do Non-Spanish-Hispanic-Latinos, as explained in 175     This study found that elderly individuals have the highest mortality rate among GBM 201 and GCG patients in comparison to adults (p<0.001). Some studies were consistent with the 202 previous findings [17][18][19][20]. Therefore, age is considered a significant predictor of survival 203 time [21]. This study also demonstrates that elderly individuals are more prone to having 204 GBM than GCG, which explains the rarity of GCG. This finding may indicate that the elderly 205 population is more susceptible to GBM due to an increased chance that cells will mutate into 206 cancer cells. The current study demonstrated that more males are afflicted with GBM than 207 with GCG, while more females are afflicted with GCG (P=0.481), consistent with [3,[22][23][24][25][26]. 208 Another study, conducted on Black patients with GBM, showed that Black males were 209 affected by GBM more than were Black females [27]. Therefore, GCG, an uncommon type 210 of glioblastoma multiform, more often affects females. However, GBM affects males more 211 than females, regardless of race. The previous findings may be explained by genetic factors. 212 diagnosed before 2010 (P<0.001). Also, one study showed that the prognosis for elderly 214 patients with glioblastoma has improved since the introduction of the Stupp regimen (i.e., 215 radiotherapy plus concomitant and adjuvant temozolomide) in 2005 [21]. This indicates that 216 year of diagnosis has a significant impact on the prognosis of glioblastoma patients. However, 217 the proportion of patients with GBM is slightly higher than the proportion of GCG patients 218 before 2010. On the other hand, the proportion of GCG incidences is slightly higher than the 219 proportion of GBM incidences after 2010 (P=0.71). 220 Patients who didn't have a Gross Total Resection (GTR) have a higher mortality rate 221 (P<0.001). Moreover, patients who hadn't undergone surgery or GTR developed GBM more 222 often than they did GCG (P<0.001). 223 Studies like [28,29] had similar findings, stating that GTR has a better survival rate than does 224 partial resection or biopsy. Brain overlap GBM and GCG tumors are associated with higher 225 mortality rates than are supratentorial and infratentorial tumors (P<0.001). This finding was 226 similar in one study [3]. 227 However, another study showed that the median survival time for both cerebellar GBM 228 (cGBM) and supratentorial GBM (sGBM) patients is eight months, though sGBM had a 229 worse prognosis as the study progressed [30]. Also, patients with brain overlap tumors have 230 a higher tendency to develop GBM than GCG (P<0.001). Because GBM is more common 231 than GCG, it affects brain overlap regions more than supra-and infratentorial regions (which 232 are affected more by GCG, P<0.001). This accounts for the higher mortality rate. Non-233 Spanish-Hispanic people have a higher mortality rate from GBM (88.6%, P<0.001). In 234 a lower incidence of GBM and present slightly younger than non-Latino Whites [31]. 236 However, white people were found to have the highest incidence of death from GBM 237 and GCG as compared to individuals of other races (P<0.001). 238 This research received no external funding. 252

Acknowledgments 253
We acknowledge Pura Rodríguez for her support in data analysis. 254

Conflicts of Interest:
255