Comparison of NCDB and CBTRUS demographic data for astrocytoma and glioblastoma

Gliomas are the most common form of malignant brain tumor in the U.S.¹ They are composed of astrocytoma (including glioblastoma), oligodendroglioma, ependymoma, oligoastrocytoma, malignant glioma, and other nonspecified and rare histologies.² Glioblastoma presents as the most common form of glioma—as well as one of the most lethal—with a five-year survival rate of 5 percent.³ This mortality rate makes assessing the treatments for glioblastoma critical. Glioblastoma incidence differs significantly within population-based analyses between age, sex, and race/ethnicity.² Reports indicate that white men have the highest rates, while incidence tends to increase with age.^1-3

A National Cancer Database (NCDB) study using 2012–2014 data showed that compared with population-based national cancer registry data, the NCDB included 85.8 percent of brain/cranial nerves malignant cases, 85.6 percent of brain and other nervous system cases, and 57.8 percent of brain cranial nerves benign/borderline cases, but did not specifically analyze astrocytoma or glioblastoma.⁴ This column compares patient demographics of astrocytomas and glioblastomas from the NCDB, a hospital-based cancer registry, and the Central Brain Tumor Registry of the U.S. (CBTRUS), a population-based registry.^2,5-6 We assessed whether the NCDB data are representative of national data for astrocytomas and glioblastomas by race, age, and sex, which are significant prognostic factors for glioblastoma.⁷ This characterization is particularly important, as NCDB data are used for analysis of treatment and survival.

Methods

NCDB and CBTRUS data on astrocytoma and glioblastoma were compared for diagnosis years 2006–2010 and 2011–2015. The International Classification of Diseases for Oncology, 3rd Edition, was used to determine primary sites (C70.0, C70.1, C70.9, C71.0–C71.9, C72.0–C72.5, C72.8–C72.9, and C75.1–C75.3) and histology/behavior codes—astrocytoma consists of pilocytic astrocytoma (9421/1) and (9425/3 only for diagnosis year group 2011–2015), diffuse astrocytoma (9400/3, 9410/3, 9411/3, and 9420/3), anaplastic astrocytoma (9401/3), and unique astrocytoma variants (9381/3, 9384/1, and 9424/3); glioblastoma consists of (9440/3, 9441/3, and 9442/3). Percentages for the subgroups of age at diagnosis, race, and sex were calculated for each cancer type, source of data, and diagnosis year. Statistical significance for comparisons within each subgroup between NCDB and CBTRUS was performed using the z-test for proportions, using the Bonferroni method of correction. Statistical analyses were completed with SAS (statistical analysis software) (9.4, Cary, NC) using the significance threshold of p <0.05.

Results

Cancer type

The percentage of astrocytomas and glioblastomas in the NCDB was similar to that in CBTRUS, with approximately 75 percent of the cases categorized as glioblastomas in each database and set of diagnosis years. Although similar, the comparison of the NCDB and CBTRUS by each cancer type for 2006–2010 and 2011–2015 showed that the differences were statistically significant (see Figure 1).

Figure 1. Percent of astrocytoma and glioblastoma by NCDB and CBTRUS in 2006–2010 and 2011–2015

Age at diagnosis comparison

Age at diagnosis was categorized as 0–14 years old, 15–39 years old, and 40 years old and older. Data were obtained from NCDB for 2006–2010 and 2011–2015. However, CBTRUS presented only 2011–2015 data because complete data from CBTRUS for this age subgroup were not available for 2006–2010.

For 2011–2015 astrocytoma data, the age distribution was similar in NCDB and CBTRUS data, with 55 percent of cases 40 years old and older in NCDB and 50 percent 40 years old and older in CBTRUS. The percentage of cases 0–14 years old were slightly lower in the NCDB compared with CBTRUS (15 percent versus 22 percent). All age differences were statistically significant between the NCDB and CBTRUS (see Figure 2).

For 2011–2015 glioblastoma data, age distributions also were similar between NCDB and CBTRUS, with about 95 percent of patients age 40 and older in both databases. No statistically significant differences were found between the NCDB and CBTRUS by age for glioblastoma (see Figure 2).

Figure 2. Percent of astrocytoma and glioblastoma by age at diagnosis for NCDB and CBTRUS in 2006–2010 and 2011–2015

Race comparison

Race was categorized as American Indian/Alaskan Native (AIAN), Asian/Pacific Islander (API), black (B), and white (W) for both sets of diagnosis years for NCDB and CBTRUS. Other and unknown race were excluded from this portion of the analysis.

For astrocytoma, the percent race distribution was similar between NCDB and CBTRUS for both 2006–2010 and 2011–2015 diagnosis years. In 2011–2015, 88 percent of NCDB and 87 percent of CBTRUS patients were white, 8 percent of NCDB patients were black versus 9 percent of CBTRUS cases, and 3 percent of NCDB patients were API versus 4 percent of CBTRUS cases. For 2006–2010 and 2011–2015, no statistically significant difference was seen between NCDB and CBTRUS for black patients, but statistically significant differences existed for all other racial subgroups (see Figure 3).

For glioblastoma cases, the race distribution also was similar between NCDB and CBTRUS. In 2011–2015, white patients comprised 92 percent of NCDB cases and 91 percent of CBTRUS cases, black patients comprised 6 percent of NCDB cases and CBTRUS cases, while API comprised 2 percent of both databases. No statistically significant difference was detected between NCDB and CBTRUS in 2006–2010 and 2011–2015 for black patients, but statistical differences for all other race subgroups were found (see Figure 3).

Figure 3. Percent of astrocytoma and glioblastoma by race for NCDB and CBTRUS in 2006–2010 and 2011–2015

Sex comparison

Sex was categorized as male and female for both sets of diagnosis years within the NCDB and CBTRUS.

For 2006–2010 and 2011–2015 astrocytoma cases, the percent distribution by sex was similar in NCDB and CBTRUS. In 2011–2015, 54 percent of cases were male in both databases. All differences in each sex between the NCDB and CBTRUS were not statistically significant (see Figure 4). For glioblastoma, the distribution by sex was similar to astrocytoma in both databases for 2006–2010 and 2011–2015. In 2011–2015, 58 percent of NCDB and CBTRUS cases were male, with all differences between the NCDB and CBTRUS for each sex not statistically significant (see Figure 4).

Figure 4. Percent of astrocytoma and glioblastoma by sex for NCDB and CBTRUS in 2006–2010 and 2011–2015

Discussion

The NCDB astrocytoma and glioblastoma comparisons indicate similarities with population-based data from CBTRUS, for diagnosis years 2006–2010 and 2011–2015. These trends were similar by cancer type, age at diagnosis, race, and sex, and in multiple instances no statistically significant difference was found between NCDB and CBTRUS within the subgroups analyzed. Analytic studies using NCDB astrocytoma and glioblastoma demographic data are thus representative of national data and could be used to identify trends in treatment, survival, and other factors.

Acknowledgment

Statistical support for this column was provided by Amanda E. Browner, MS, Statistician, NCDB.