DISCUSSION
Overall, non-Hispanic whites had the highest 10-year survival, with
non-Hispanic blacks, Hispanics, and Other having a worse survival.
Considering histologic types, anaplastic astrocytoma and glioblastoma
had the lowest survival and diffuse astrocytoma (protoplasma,
fibrillary) and mixed glioma had the highest survival. This is
consistent with findings from previous
studies.8,28,29,30,31,32 This study differs from other
studies of pediatric brain and CNS tumor survival because it assesses
the racial disparities in brain and CNS tumor survival using recent
national specialized census tract data that provide information on both
SES and cancer treatment received by patients.
The largest group of minority patients were Hispanics, which is not
surprising because Hispanics are currently the largest racial minority
and one of the fastest growing racial minority groups in the
US.33 Similar to what was found in other
studies6,11,12 there was a male predominance in
diagnosed cases and the majority of cases were diagnosed before age ten.
Pilocytic astrocytoma was found to be the most common tumor type in
children, with non-Hispanic whites having the highest proportion of
cases. This concurs with findings from other
studies.6,12 It was also previously shown that
non-Hispanic blacks had the highest burden of
medulloblastoma,34 however, similar to the findings of
Cooney, et al. and Barnholtz-Sloan et al.,7,11 our
study showed Hispanics had the highest burden of both medulloblastoma
and ependymoma. “Other” race/ethnicities had the highest proportion of
anaplastic astrocytoma and glioblastoma histology types. Studies that
examined these histologies separately found that non-Hispanic whites had
an increased proportion of glioblastoma tumors8 while
non-Hispanic blacks had an increased proportion of anaplastic
astrocytoma.11
Even though our study showed a strong association between race and SES,
similar to Austin and Kehm’s findings,6,22 SES was not
significantly associated with adverse cancer outcomes in the univariate
and multivariate models for both 5-year and 10-year risk of death.
Accounting for SES in the multivariate model reduced the risk of death
across all racial groups. This supports findings from previous studies
on the relationship between SES and solid tumor outcomes demonstrating
that low SES is not associated with poor survival
outcomes.6,22 Results from our study suggest that
factors other than SES mediate the racial survival disparities observed
in brain and CNS tumors.
Examining the multivariate analysis, we found that the 10-year hazard of
death was significantly increased for all three minority racial groups
as compared to non-Hispanic whites, with Hispanics having the highest
risk of death. Survival reaches a plateau for most tumors quite late
with the exception of high-grade gliomas and
medulloblastoma.35,36,37 It is thus not surprising
that in our study, the racial/ethnic survival disparities were more
pronounced in the 10-year models. Late mortality could also occur in
5-year survivors, attributable to multiple causes including late
treatment effects, tumor recurrence or progression which could be
influenced by quality of initial therapy.38,39,40 It
will be informative to look at the cause-specific mortality for cases.
The relatively poor 5-year survival seen among Hispanics could be due to
multiple factors. Based on the study results, Hispanics have the highest
proportion of patients undergoing partial lobectomy. This procedure was
associated with an increased 5-year risk of death based on the study
results. Similar to finding from other studies6,41,
our results showed that Hispanics have a slighter higher burden of cases
reporting with advanced cancer compared to the other racial groups. It
may stem from the fact that the affected individuals may be
uninsured/underinsured which is influenced by several factors including
parents’ immigration status and healthy families who choose to forgo
insurance.42,43
In our study, we found significant differences in treatment modality by
race/ethnicity, which may partially explain the increased risk of death
seen in minority racial groups. Similar treatment differences were noted
in a study which addressed the disparities in brain tumor
treatment.13 Considering that treatment received is
known to impact survival,19 patients who receive
inadequate treatment or poor quality treatment have less chances of
survival. Minority populations as compared to non-Hispanic whites often
receive treatment from low volume hospitals which are less equipped to
handle such cases rather than high volume hospitals or specialized
cancer centers where they can receive adequate and quality care from a
multidisciplinary clinical team for both initial therapy and follow up
care for survivors.10,13,38 Furthermore, there is
evidence to suggest that racial discrimination on the part of healthcare
providers, minority patients being less likely to be enrolled in
clinical trials, and patients declining treatment options due to
mistrust of the health system contribute to disparities in
treatment.13,20,44,45,46,47,48 There could also be
racial differences in the response to chemotherapeutics due to potential
differences in pharmacogenomics and tumor molecular markers across the
racial/ethnic groups.49,50 However, there doesn’t
appear to be a documentation of raced-based biological differences in
these factors and the extent to which they impact treatment
outcomes.37 Future studies should specifically explore
host and tumor genetic factors and its influence on treatment response
across racial groups.
Challenges with patient provider communication especially when it
involves patients with limited English proficiency (LEP), the majority
of which are from minority racial groups51 could
negatively impact delivery of optimum health care, patients’ compliance
to therapy, follow up care and ultimately treatment outcomes. Few
studies done to evaluate the impact of language barrier on cancer
treatment outcomes indicate that there is limited information shared
between providers and patients even when interpreters are engaged due to
difficulties in direct communication between the two
parties.52,53,54 The impact of patient and provider
communication and the role of medical interpreters in cancer treatment
and outcomes need to be explored further in future studies.
The study has several limitations. Data for this study were obtained
from SEER and the supplementary Specialized Census Tract-level SES and
Rurality Database which provides information on the overall SES of the
census tract of residence of a patient and not the SES of the
patient.25,26 Only information on the insurance status
of patients diagnosed from 2007 – 2015 was available, hence the
influence of insurance on treatment outcome was not assessed in this
study. Additionally, Asians, American Indians, and cases without racial
identity were grouped together due to their small sample sizes, which
does not allow for good representation of the independent survival
outcomes of Asians and American Indians. Treatment information in SEER
is incomplete with no clear differentiation between cases who did not
receive treatment and those whose treatment information is missing,
making it difficult to infer real differences in treatment among the
racial groups. SEER does not have information on molecular tumor markers
or genomic ancestry which may influence the racial disparities in
outcome.55,56,57 Finally, the data do not include
patients’ choice in the treatment.