Area-level socioeconomic status is positively correlated with glioblastoma incidence and prognosis in the United States

In the United States, an individual’s access to resources, insurance status, and wealth are critical social determinants that affect both the risk and outcomes of many diseases. One disease for which the correlation with socioeconomic status (SES) is less well-characterized is glioblastoma (GBM), a devastating brain malignancy. The aim of this study was to review the current literature characterizing the relationship between area-level SES and both GBM incidence and prognosis in the United States. A query of multiple databases was performed to identify the existing data on SES and GBM incidence or prognosis. Papers were filtered by relevant terms and topics. A narrative review was then constructed to summarize the current body of knowledge on this topic. We obtained a total of three papers that analyze SES and GBM incidence, which all report a positive correlation between area-level SES and GBM incidence. In addition, we found 14 papers that focus on SES and GBM prognosis, either overall survival or GBM-specific survival. Those studies that analyze data from greater than 1,530 patients report a positive correlation between area-level SES and individual prognosis, while those with smaller study populations report no significant relationship. Our report underlines the strong association between SES and GBM incidence and highlights the need for large study populations to assess SES and GBM prognosis to ideally guide interventions that improve outcomes. Further studies are needed to determine underlying socio-economic stresses on GBM risk and outcomes to identify opportunities for intervention.


Introduction
Glioblastoma (GBM) is a grade IV glioma that is the most common malignant primary brain tumor, comprising 60% of all cases with an annual incidence of 5.0 per 100,000 (1). According to the Central Brain Tumor Registry of the United States (CBTRUS), over 63,000 Americans were diagnosed with GBM between 2015 and 2019 (2). Standard of care currently involves surgical resection followed by radiation therapy and chemotherapy, but even with this aggressive treatment, the overall five-year survival rate is only 5% (3). Many efforts have been made over the years to better elucidate the underlying biological and physiological basis of GBM, but the socioeconomic factors underlying disease development and outcomes have received less attention. When considering the United States, where access to quality healthcare largely depends on an individual's economic resources, such as income and health insurance, socioeconomic factors are particularly crucial to analyze with respect to the incidence and prognosis of any disease, including GBM. Previous studies have uncovered correlations between area-level socioeconomic metrics and either GBM or other gliomas (4)(5)(6)(7) in the United States on the regional, state, and national levels. Here, we report a comprehensive review to synthesize current knowledge regarding the correlations between socioeconomic status (SES) and GBM incidence and prognosis in the United States.

Methods
A review of studies was conducted by querying those published between January 1, 1990, and June 1, 2022, on Google Scholar (Google Scholar, RRID : SCR_008878), PubMed (PubMed, RRID : SCR_004846), Scopus (Scopus, RRID : SCR_022559), and Web of Science (Web of Science, RRID : SCR_022706). The Cochrane database (Cochrane Library, RRID : SCR_013000) was searched to determine that no identical review had previously been conducted. The terms used for these searches were "socioeconomic status," "SES," "income," "glioblastoma," and "GBM." All studies found using this method were initially recorded, then further filtered. The criteria for selection were studies that analyzed patients in the United States and characterized a relationship between SES and either GBM incidence, GBM-specific survival, or overall survival (OS) in patients with GBM. SES involved either a measurement of income itself, an aggregate index for SES involving income, defined economic variables (such as education, poverty, unemployment, rent, and house value measures), or a selfidentified SES measurement whose components were not specified. SES was also measured on either the individual or area level. Use of these search criteria led to identification of 80 studies. These studies were then individually reviewed by full-text analysis and synthesized in a narrative review. Studies were excluded if the target population was outside of the United States, if the cancer analyzed was not GBM, if SES was not included as a variable, or if the outcome measured was not GBM incidence or disease-specific or overall survival of GBM patients ( Figure 1). The full-text analysis yielded a final list of 16 studies (three for GBM incidence and 14 for GBM prognosis, including one study that measured both) that were included in the Results. The following characteristics were recorded: title, year, authors, participants, data sources, SES measure, methodology, dependent outcomes, and key findings.

Results
Overall, 16 studies met the criteria for inclusion: two for SES and GBM incidence, 13 for SES and GBM specific survival or OS, and one for SES and both GBM incidence and OS. The GBM incidence papers analyzed between 3,832 and 45,696 patients and were published between 2005 and 2019 ( Table 1). The GBM survival and OS papers analyzed between 116 and 61,346 patients and were published between 2007 and 2022; 12 papers measured OS as their outcome, one measured GBM-specific mortality, and one measured both ( Table 2). All papers utilized area-level, not individual-level, SES measures; three examined the county level, four at the zip code Identification, Screening, and Selection of Studies for Review.   level, six at the census tract level, one at the address level using a pre-existing index, and three were not specified. SES was measured solely by income in eight of the paperseither average, median, or quintile/quartile/tertile. Six papers utilized an aggregate SES metric involving income as well as other variables, and two papers did not specify their SES measurement. All three papers that correlated SES to incidence of GBM found a greater incidence of GBM in higher SES regions (Table 1). Of the 14 papers that correlated SES to patient survival, the eight with greater than 1,530 patients analyzed found that patients living in regions with higher SES experienced longer survival on average. The six studies with ≤ 1,530 patients found no significant relationship between area-level SES and survival in GBM patients, highlighting the need for larger patient studies ( Table 2). With respect to patient population, nine papers analyzed SES and either GBM incidence or prognosis from national databases.
Specifically, six papers obtained GBM patient information from the Surveillance, Epidemiology, and End Results Program (SEER), three papers acquired data from the National Cancer Database (NCDB), and one paper included incidence data from the CBTRUS. Another five papers obtained data from university hospital records, and the two remaining papers analyzed GBM records from state-or citylevel databases. GBM incidence studies, in aggregate, contained data for patients diagnosed between 1974 and 2015, while GBM prognosis studies included patient data between 1981 and 2017.

Discussion
This review uncovered a positive correlation between both arealevel SES and GBM incidence, as well as area-level SES and survival in GBM patients in studies with larger sample sizes. When considering these results, it is important to keep in mind that SES is measured at the level of a geographic region, so it cannot be applied to individual-level SES associations with GBM incidence or prognosis. Individuals who live in an area with a higher SES experience a higher rate of GBM incidence, but an individual with a higher SES does not necessarily display a greater risk of developing GBM. When considering the positive association between SES and GBM incidence, it is important to examine the interface between these variables and age distribution. Higher SES areas tend to have a greater life expectancy (24), and the average age of onset for GBM is 64 years of age (25). However, all three papers that examined this relationship utilized age-adjusted GBM incidence (8-10), so age differences between high and low SES regions should not explain this relationship. Another important consideration is race and ethnicity, as there is ample evidence that non-Hispanic whites are the most at-risk for developing GBM (2,10), and that this group is over-represented in high SES regions. However, two of the papers analyzing SES and GBM incidence controlled for race in their incidence calculations (8,10), and the third stratified incidence calculations by race, demonstrating that non-Hispanic white, white Hispanic, and black patients experience greater GBM incidence as area-level SES increases (9). Additionally, one may consider that patients in high SES regions may have greater access to diagnostic modalities that reveal GBM, artificially increasing GBM diagnoses in these regions. Clinical suspicion for GBM begins with apparent space-occupying lesions on computer tomography or magnetic resonance imaging, and GBM typically progresses to severe signs and symptoms, such as seizure, even if initial presentation was nonspecific (26). Therefore, it is unlikely that many patients would progress through their entire disease course without proper diagnosis either pre-or post-mortem. Overall, it remains unclear why there is a direct relationship between SES and GBM incidence.
The positive association between SES and survival in GBM patients, on the other hand, was only observed in studies with larger sample sizes. This is likely due to the variability in metrics used to assess SES and the heterogeneity in SES at different geographical (e.g., census tract, zip code, versus county) regions. This association could be explained by a multitude of factors, the first being type of health care insurance. Insured GBM patients have a better prognosis than those uninsured (27), patients with private insurance survive longer than those on Medicaid (28), and non-Medicaid patients survive longer than those on Medicaid or who are uninsured (29). Therefore, perhaps individuals in high SES areas are more likely to have private insurance, leading to better access to care and therefore better prognosis.
Another factor explaining this relationship may be the type of treatment received. GBM patients living in higher SES regions have a greater chance of receiving radiation, and those who receive this treatment have greater OS (30)(31)(32). The same can be said for GBM patients who receive triple therapy (surgery, radiation, and chemotherapy) (33,34). An additional consideration is clinical trial participation: those living in greater SES regions are more likely to participate in clinical trials (23), which is correlated with improved OS (35) and allows for more salvage therapy options. Interestingly, it does not appear that Karnofsky Performance Status differs between those living in high and low SES regions (19,23). Therefore, this relationship cannot be explained by patients in low SES regions presenting with more severe disease than those in high SES regions and may instead reflect differences in access to quality care.
There are several limitations to the conclusions that can be drawn from this review. First, due to the ecological fallacy, interpretation must remain on the area level and cannot be applied to individuals with high or low SES. Second, this study is not a systematic review or meta-analysis, as it was not anticipated that there would be many studies that would qualify for quantitative analysis. Consequently, it is possible that additional qualifying studies exist that were missed by the identification process. Third, while there was no overlap in patient data in Table 1 (SES and GBM incidence), there is likely substantial patient overlap in Table 2 (SES and GBM prognosis), as multiple studies included either SEER or NCDB data in varying time periods that overlapped between 2005 and 2010. Therefore, there are likely redundant conclusions pulled from approximately half of these studies (including 10,12,14,17,18,20,22). It is critical to note that while there appears to be a positive correlation between area-level SES and GBM prognosis, several of the studies characterizing this relationship analyze many of the same patients. Finally, due to publication bias, it is possible that more studies than those included here were conducted that found no association between area-level SES and GBM incidence or survival but remain unpublished due to negative results.
Placing these results in a broader context, the relationship between area-level SES and cancer incidence appears to depend on the cancer site in question. Past studies reveal that the risk of gastric, colorectal, larynx, cervix, penile, and liver cancer are greater in low SES regions, whereas the risk of melanoma, thyroid, and testicular cancer are greater in high SES regions (36)(37)(38)(39). On the other hand, it appears that there is a positive correlation between cancer prognosis and area-level SES across cancer type, including breast cancer (40), Hodgkin lymphoma (41), non-small cell lung cancer (42), liver, kidney, colorectal, and prostate cancers (43), and various childhood cancers (44). Therefore, the relationship between SES and GBM prognosis is comparable to other cancer sites.
In this review, both greater GBM incidence and survival of GBM patients in high SES regions were observed, specifically in studies with larger sample sizes. As research continues to be conducted on the social determinants of health related to GBM, it will be important to utilize these findings to improve patient outreach, clinical trial enrollment, and education in those areas where patients are more likely to develop GBM or exhibit worse prognosis from the disease. Hopefully, as treatments for GBM become more effective, such interventions will reduce disparities in health care and outcomes in those living in regions of varying SES.