Identifying disparities in care in treating glioblastoma: A retrospective cohort study of patients treated at a safety-net versus private hospital setting.

Abstract

Background

Patients of lower socioeconomic status (SES) may experience barriers to their oncologic care, but current data is conflicted over whether SES affects the prognosis of patients with glioblastoma (GB).

Objective

We sought to determine whether SES disparities impaired delivery of neuro-oncologic care and affected the prognosis of GB patients.

Methods

The records of GB patients treated from 2010–2014 at a safety-net hospital (SNH) or private hospital (PH), both served by one academic medical institution, were retrospectively reviewed and compared. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan–Meier method.

Results

55 SNH and 39 PH GB patients were analyzed with median 11 month follow-up. SNH patients were predominantly Hispanic, low income, and enrolled on Medicaid, SNH patients were less likely to receive radiation (89% vs. 100%), took longer to start radiation (41 vs. 29 days), and were less likely to complete radiation treatment (80% vs. 95%). Concurrent and adjuvant temozolomide use were also lower (85% vs. 94% and 60% vs. 71%, respectively). OS and PFS were not significantly different (15 vs. 16 months and 8 vs. 11 months, respectively). On multivariate analysis, adjuvant chemotherapy and RT completion predicted for better OS while hospital type, income, and insurance did not.

Conclusion

Although GB patients at our SNH received less adjuvant treatment compared to PH, outcomes were similar. Access to multidisciplinary care staffed by academic physicians may play an important role in overcoming socioeconomic barriers to treatment availability and quality at SNHs.

Introduction

The most common primary brain tumor in adults, glioblastoma (GB) is an aggressive neoplastic disease that is almost uniformly fatal despite aggressive multidisciplinary care. The current standard of care involves maximal safe surgical resection followed by high dose radiotherapy with concurrent temozolomide (TMZ) with an additional course of adjuvant chemotherapy.¹ Enrollment in clinical trials have been demonstrated to improve outcomes.² Nevertheless, the majority of GB patients die of their disease within two years. Recurrence can be rapid, requiring frequent imaging to monitor for progression of disease. Specific subsets of patients, including the elderly, the uninsured, and patients with low Karnofsky performance scores (KPS) have unfavorable prognoses.³

Patients of lower socioeconomic status (SES) may experience barriers to their oncologic care, which lead to delayed presentation and treatment. Understanding potential disparities in care has implications for treating GB patients by identifying interventions that optimize outcomes. It is believed that SES can impact outcomes in a variety of other malignancies, but the current data is conflicted over whether SES affects the prognosis of patients with GB.^4,5 Some studies indicate that prognosis is not dependent on SES.^5,6 This has been attributed to similar treatment despite lower SES, such as similar rates of surgical resection.⁷ Other studies demonstrate that socioeconomic barriers to care influenced the performance status of patients treated at public hospital for GB; which, in turn, impacted their survival.^4,8,9

GB requires complex multidisciplinary care due its aggressive nature and need for trimodality treatment. Unlike private hospitals (PH), public or safety net hospitals (SNH) serve a predominantly Medicaid and uninsured population. GB patients seen at a private institutions may receive more treatment and ultimately have better outcomes compared to those at a public hospital because of socioeconomic barriers leading to different patterns of care.^10,11 A recent study by Mandel et al. found that though patients treated at SNH had worse outcomes compared to those at a PH, but similar OS was achieved when SNH patients received standard of care treatment with adjuvant chemoradiation.⁹ For this retrospective study, we examined patients seen at one public SNH and one PH staffed by the same physicians to see if there were differences in GB care and ultimately prognosis.

Materials and Methods.

Study Population

The institutional review board of the *** (ID:HS-13–00802) waived the need for ethics approval and the need to obtain consent for the collection, analysis and publication of the retrospectively obtained data for this non-interventional study under the *** Human Research Protection Program Flexibility Policy. Using institutional treatment databases, we reviewed records of patients treated for GB at the *** (*** PH) or the *** (***; SNH) from 2010 to 2014. Both *** and *** are *** teaching hospitals, but each hospital has a separate administration. As a SNH, *** primarily accepts patients who are uninsured or have Medicaid. In all cases, a pathologist confirmed the diagnosis of GB using standard diagnostic classification systems (World Health Organization (WHO) Classification System). The same group of neurosurgeons, radiation oncologists, and neuro-oncologists provide multi-disciplinary care to patients at both hospitals. To better compare pattern of cares and possible outcomes between the two hospital systems, patients who received the majority of their initial oncologic care at either *** or *** were selected. This was defined as patients who underwent surgical resection at either hospital who received one element of their adjuvant treatment (chemotherapy or radiation) at the same center or their entire adjuvant chemoradiation course at one hospital after surgical resection at an outside hospital. In addition, patients who received surgical resection at a hospital but did not receive adjuvant treatment were included. Patients who were only treated for recurrent disease or were documented to have received adjuvant chemoradiation at an outside hospital system were excluded from the analysis. Patients without adequate follow-up (< 6 weeks), outcome data, operative reports, or hospitalization information were excluded.

Data Collection

We reviewed medical records to obtain patient demographic information, including age, race, gender, insurance status, and residential zip code. Household incomes were estimated by using aggregate zip code data from the 2016 American Community Survey. Low SES was defined as living in a zip code where the average income was less than $50,000 a year, which is approximately 200% of the 2016 federal poverty level for a family of four. Tumor characteristics included tumor location, tumor size, clinical presentation, performance status, IDH mutation status, and O6-methylguanine DNA methyltransferase (MGMT) methylation. Maximum diameters were obtained from the medical record or were measured manually in the institutional picture archiving and communication system. Recursive Partitioning Analysis (RPA) classification was determined retrospectively from the patient record. Treatment characteristics such as extent of surgery, radiation technique dose and fractionation, receipt of concurrent and adjuvant chemotherapy, clinical trial enrollment, and salvage therapy were assessed. The surgical operative note and/or radiological report was used to determine if gross total resection likely achieved, and otherwise was considered to be a subtotal resection. Toxicity during adjuvant chemoradiation was graded with the Common Terminology Criteria for Adverse Events (version 4.03). The date of last follow-up was the last clinical encounter documented in medical records. Survival data were obtained from institutional cancer registries and public online databases. For patients with adequate follow-up in the clinic, progression-free survival (PFS) was analyzed.

Data Analysis

Baseline patient information, treatment characteristics, and rates of salvage treatments were compared with the Mann-Whitney U-test rank-sum test for continuous variables and the Pearson chi-square test for categorical variables.

Overall survival (OS) and PFS were calculated using the Kaplan-Meier method. Statistically significant comparisons between the two hospitals were calculated with the log-rank test. Univariate and multivariate analyses were performed with the Cox proportional hazards model. All risk factors, including the institution, age, race, income, marital status, insurance, tumor size > 5cm, unifocal versus multifocal tumor, extent of resection, receipt and completion of radiation, time to RT, concurrent chemotherapy, and adjuvant chemotherapy were entered into the univariate analysis with each clinical outcome. Risk factors with P values < 0.10 were further entered into the multivariate analysis. Statistically insignificant variables were removed from the multivariate analysis in a stepwise fashion until only significant variables remained. Hazard ratios associated with covariates were estimated. All analysis was performed with JMP Pro (SAS Institute Inc. Cary, NC). For all statistical testing, P < .05 was considered significant. This article adhered to the reporting guideline of Strengthening the Reporting of Observational studies in Epidemiology for cohort studies.

Results

Patient Characteristics

A total of 116 patients were queried from our institutional records and retrospectively reviewed in our study. Twenty-two patients were excluded from analysis for a variety of reasons: 13 patients received their entire adjuvant oncologic care at outside centers, one was treated at both the SNH and PH, three had previously treated gliomas that underwent malignant transformation, and five patients had inadequate follow-up. The remaining 94 patients, 55 from the SNH and 39 from the PH, were included and analyzed.

In the entire cohort, the median age was 54 years (Interquartile range [IQR], 47–61 years), 40 patients (42%) were female, and 38 patients (40%) were non-Hispanic White. SNH patients were predominantly Hispanic (75% vs. 18%, P<0.0001), younger (53 vs. 59 years, P=0.0056), and unmarried (51% vs. 13%, P=0.0004) compared to their PH counterparts. Patients at the SNH were more likely to be low income (56% vs. 26%, P=0.0027) and have Medicaid or uninsured (100% vs. 3%, P<0.0001). There was no statistical difference in gender between institutions.

Two measures of tumor characteristics, performance status and RPA classification, did not differ between patients between hospitals. Multifocal disease at presentation was uncommon in general and did not vary between cohorts. Table 1 summarizes baseline patient characteristics.

Table 1.

Safety Net (***) and Private (***) Hospital Patient Characteristics

		SNH (n=55)	PH (n=39)
Age	Median (IQR)	53 (46–57)	59 (50–68)	P=0.0056
Ethnicity				P<0.0001
	White	10	28
	Black	0	1
	Hispanic	41	7
	Asian	4	3
	Other	0	0
Gender				P=0.6530
	Male	31	23
	Female	24	16
Marital Status
	Married	28	34	P=0.0004
	Single/Divorced	27	5
Income				P=0.0027
	<50K	31	10
	>50K	24	29
Insurance				P<0.0001
	Private	0	28
	Medicare	0	7
	Medicaid	52	1
	Uninsured	3	0
KPS				P=0.1151
	90–100	14	16
	70–80	27	16
	50–60	9	6
	30–40	2	1
	10–20	3	0
Time from Reported Symptoms to Presentation
	Weeks (IQR)	4 (2–7)	4 (2–8)	P=0.6252
Tumor Size				P=0.4310
	>5cm	28	17
	<5cm	26	22
Tumor Site				P=0.2530
	>1	10	11
	1	45	28
IDH				P<0.0001
	Wild Type	0	14
	Mutated	0	2
	N/A	55	23
MGMT
	Methylated	0	2	P<0.0001
	Unmethylated	0	9
	N/A	55	28

Treatment Characteristics

There was non-significant trend towards a difference in rates of patients who had gross resection of the enhancing tumor at the SNH versus the PH (20% vs 39%, P=0.0857), but with similar complications rates and length of hospitalization. However, there were disparities in delivery of adjuvant therapy at the SNH. SNH patients were less likely to receive radiation (89% vs. 100%, P<0.0001), took longer to start radiation (41 vs. 29 days, P=0.0002), and were less likely to complete radiation treatment (80% vs. 95%, P=0.0297). Patients treated at the SNH were less likely to receive concurrent TMZ and adjuvant TMZ (85% vs. 94% patients, P=0.0021 and 60% vs. 71% patients, P=0.021 respectively). Clinical trial enrollment was lower at SNH (15% vs. 33%, P=0.03). Table 2 summarizes treatment details by site.

Table 2.

Treatment Characteristics

Surgery Details		SNH	PH	P Value
Surgery Center
	SNH	47	0
	PH	0	37
	OSH	8	2
Surgical Resection				0.0857
	Gross Total	11	14
	Subtotal	44	25
Post-Op Complications		6	3	0.5975
Length of Hospitalization
	Median Days (IQR)	3 (2–7)	2 (1.5–5)	0.1052
Radiation Details
Radiation				0.0095
	Yes	50	39
	No	5	0
Radiation Site
	SNH	50	0
	PH	0	30
	OSH	0	9
RT Modality				0.0448
	IMRT	40	33
	3DRT	6	0
	Proton	0	1
	N/A	4	5
RT dose received	Median Gy (IQR)	59.4 (59.4–60)	59.4 (59.4–60)	0.1632
Time to RT Start	Median Days (IQR)	41 (29.25–53.5)	29 (24–36)	0.0002
RT elapsed days	Median Days (IQR)	43 (41.5–47)	(44.5 (41–49)	0.4742
Treatment Delay		10	3	0.0828
RT Completed?		44	37	0.0297
Progression During RT		4	2	0.6639
Chemo Stopped During RT		4	2	0.558
Chemotherapy Details
Clinical Trial				0.0300
	Yes	8	13
	No	47	26
Concurrent Chemo				0.0021
	Temozolomide	47	37
	TMZ + Avastin	0	2
	None	8	0
Adjuvant Chemo				<0.0001
	Temozolomide	33	28
	Avastin	0	2
	TMZ+Avastin	0	5
	None	23	4
Adjuvant Chemo Completed		8	8	0.5978
Reason for Stopping Chemo				0.1530
Progression		29	22
Comorbidity		2	3
Toxicity		1	4
Patient Decision		5	1
N/A		10	1

Patient Outcomes

The median follow up time of was 11 months (IQR 6–23 months) for all patients. The median OS for patients treated at either the SNH and PH was similar at 15 versus 16 months respectively (P=0.4877, Figure 1). The median PFS for SNH and PH patients was not statistically significantly different (8 vs. 11 months, p=0.28, Figure 2). The two and five year OS was 33% and 12% for patients who were treated at a SNH and 22% and 3% for patients who were treated at a PH. Rates of salvage surgery and salvage radiation were not statistically significantly different (Table 3).

Figure 1.

Kaplan Meier curve estimating overall survival at a SNH (***) and PH (***). The median OS for patients treated at either the SNH and PH was similar at 15 versus 16 months respectively (P=0.4877).

Figure 2.

Kaplan Meier curve estimating progression free survival at a SNH (***) and PH (***). The median PFS for SNH and PH patients was similar (8 vs. 11 months, p=0.28).

Table 3.

Characteristics of patient follow up after RT and rates of salvage treatment.

Outcome Details		SNH (n=55)	PH (n=39)	P Value
Time to 1st MRI	days	41.5 (17–66)	17 (10–26)	0.0005
Time Between MRIs	months	2.2 (1.6–3.2)	1.5 (1.15–1.88)	0.001
Hospitalization due to Progression		14	9	0.3492
Salvage RT				0.0629
	Yes	0	2
	No	44	31
Salvage Surgery				0.5037
	Yes	10	10
	No	34	24
Time to 1st MRI	days	41.5 (17–66)	17 (10–26)	0.0005
Time Between MRIs	months	2.2 (1.6–3.2)	1.5 (1.15–1.88)	0.001

Table 4 summarizes toxicity during adjuvant treatment. PH patients were more likely to report fatigue (18% vs. 43%, P=0.0073). Otherwise, rates of acute toxicity were similar between the two patient cohorts. The most common acute G1–2 side effects were alopecia (32% vs. 28%, P=0.64) and headache (20% vs. 10%, P=0.2037). Only 5% and 7% of patient experience G3+ thrombocytopenia at the SNH and PH respectively (P=0.6619). No other severe toxicities were seen.

Table 4.

Acute toxicities during treatment.

Toxicity		SNH (n=55)	PH (n=39)	P Value
Fatigue
	G1–2	10	17	0.0073
Dermatitis
	G1–2	7	7	0.4835
Alopecia
	G1–2	18	11	0.64
Headache
	G1–2	11	4	0.2037
Seizure
	G1–2	0	1
Vision Change
	G1–2	2	1	0.7707
Dry Mouth
	G1–2	0	1
Nausea
	G1–2	9	2	0.095
Constipation
	G1–2	4	3	0.9391
Thrombocytopenia
	G1–2	2	5	0.0947
	G3+	3	3	0.6619

On multivariate analysis (Table 5), failure to receive adjuvant chemotherapy and failure to complete RT were associated with worse OS (HR 4.017 P<0.0001 and HR 3.1682 P=0.0427 respectively). Hospital type, income, and insurance status were not associated with OS on multivariable analysis. Only the lack of concurrent or adjuvant chemotherapy was associated with worse PFS (HR 9.405, P=0.0246 and HR 3.116, P=0.0084 respectively).

Table 5.

Univariate and Multivariate Cox proportional hazards survival analysis of factors associated with survival in patients with GB

		Univariate Analysis		Multivariate Analysis
Parameters		HR	P Value	HR	P Value
Age >50		1.85	0.0113	1.406	0.2114
Ethnicity
	Asian: White	1.21	0.675
	Hispanic: White	0.8907	0.9078
	Black: White	0.9375	0.7854
Gender	Female	1.3615	0.1845
Marital Status	Yes	1.2279	0.3981
Income	>50K	1.3103	0.2487
Insurance
	Medicare:Private	0.6943	0.3798
	Medi-Cal: Private	0.7791	0.3121
Tumor Size	>5cm	0.9701	0.8963
Tumor Site	>1	0.7	0.2245
KPS
	70–80	1.6108	0.0726	1.2804	0.3993
	<70	4.0395	0.0001	1.7603	0.145
Surgical Resection	Subtotal	1.5482	0.0778
Radiation	No	16.7244	<0.0001	1.607	0.6095
Clinical Trial	Yes	0.5724	0.0312	0.8845	0.6807
Concurrent TMZ	No	10.5077	<0.0001	3.2142	0.2008
Adjuvant TMZ	No	6.229	<0.0001	4.017	<0.0001
Site	PH:SNH	1.1675	0.4981
Time to RT		0.9957	4137
Elapsed Days of RT		0.9739	0.0687
RT Complete	No	7.3354	<0.0001	3.1682	0.0427

Discussion

The goal of our study was to examine the effects of socioeconomic disparities and hospital subtype on patients treated for GB. Prior studies have suggested patient socioeconomic characteristics including race, income, and insurance status can impact the delivery of treatment and oncologic outcomes. However, patients at SNHs who received standard of care treatment can achieve similar outcomes. We conducted our study jointly at two hospitals located in ***: the ***, a tertiary academic medical center, and ***, one of the largest SNHs in the United States. Given the unique affiliation between the SNH and the PH as teaching hospitals with the same physician staffing, we were optimally poised to isolate the impact of socioeconomic disparities and hospital subtype on patients with GB.

Poorer survival in our study was associated with lack of adjuvant chemotherapy and inability to complete radiation therapy. Stupp et al demonstrated the benefit of concurrent radiation treatment with temozolomide in a phase III trial.¹ Patients at our SNH were less likely to receive chemotherapy and radiation after their surgery or were unable complete their adjuvant course. Though lack of adjuvant chemotherapy and inability to complete radiation could be an indicator of progression during chemoradiation, we found similar rates of progression between institutions during treatment. Therefore, it is more likely that other factors contributed to this discrepancy in completing adjuvant treatment. There was a non-significant trend towards a greater proportion of SNH patients presenting with poor performance status (25% vs. 18%) compared to PH patients, which may affect whether they are offered chemoradiation and complete treatment. It is possible that SNH patients may be less informed about their course of care or face barriers that delay or prevent follow-up for adjuvant therapy.¹² In addition, patients treated at the SNH took longer to start radiation after their surgery in our study, which was concerning because in many cases SNH patients face barriers to follow up at post-surgical appointments. In a large retrospective study, patients with Medicaid, other government insurance, or those who are uninsured have higher odds of receiving radiation >35 days after gross total resection.¹³ In a meta-analysis of 19 studies examining the impact of delay on survival, Loureiro et al. found no evidence that delay of radiation affected OS in patients with GB.¹² This mirrors our findings that this delay did not translate into a survival detriment.

In our analysis, despite expectations, it was reassuring that hospital type, income, ethnicity, insurance, and clinical trial enrollment were not associated with PFS or OS. Patients treated at our SNH were predominantly Hispanic and had Medicaid, while the majority of the patients treated at our PH were non-Hispanic White and had private insurance. Other retrospective studies have reported the effect of racial and ethnic disparities on outcomes for GB patients. Similar to our study, two single institution studies and a SEER database review suggesting similar OS for Hispanic patients compared to non-Hispanic Whites^14,15 However, despite similar survival, several SEER analyses having found disparities in delivery of care in Hispanic patients, who were less likely to receive radiation therapy and had greater rates of sub-total resection compared to non-Hispanic Whites.^16–18 Difference in delivery of may be due in part to non-English speaking patients experiencing barriers to effective communication of the plan of care with their physician. Socioeconomic factors, such as lower income and education, may also be associated with both non-White ethnicity and omission of RT.

Insurance status can affect the rates of receiving or completing standard of care adjuvant therapy. At our SNH, 95% of patients were enrolled in Medicaid, while the remaining patients were uninsured. Two single institution studies found that patients with private insurance were more likely to receive concurrent chemotherapy compared to those with Medicaid, Medicare, or uninsured, while another study found the same trend for radiation.^19–21 Two large database studies found in their analysis that patients with private insurance had better survival than those with Medicare or who were uninsured.^10,22 Mandel et al. found that the vast majority of their SNH patients (71.8%) had no insurance and that lack of insurance was associated with worse OS.⁹ Only 45.8% of their SNH patients received standard of care with radiation and TMZ, which was attributed partially to patients lacking insurance. This is similar to a recent study that found one that reported a significantly lower rate of postoperative treatment with radiation and TMZ use in uninsured (56.3% and 56.3%, respectively) and Medicaid patients (66.7% and 64.4%, respectively).²⁰ Comparatively, we found no disparity in PFS between patients treated at our SNH versus the PH. The similar survival and progression outcomes despite disparities delivery of care could be because, in part, the majority of patients at the SNH received standard of care treatment (concurrent chemoradiation 85% and adjuvant chemotherapy 60%) which was similar to our patients with private insurance (95% and 72% respectively). Compared to our SNH, a large database study found lower the rates of postoperative treatment with radiation and TMZ in uninsured (47% and 41% respectively) and Medicaid patients (62% and 53% respectively).²³ These numbers were similar to a recent study using patients from the National Cancer Database, which suggested lower rates of gross total resections (35.2 vs 38.0%), radiation (73.4% vs 79.7%), and chemotherapy (69.9% and 76.%) at high Medicare/Medicaid burden hospitals compared to those that treated patients mainly with private insurance, which the authors hypothesize was the contributing factor to lower 30 and 90 day survival. We hypothesize that despite socioeconomic barrier, there is a higher utilization of adjuvant radiation and TMZ at our SNH due in part because both the SNH and PH are medical centers staffed by academic physicians. Treatment at academic centers has been shown to improve overall survival in a large database analysis.^10,23 Since academic physicians at our PH also staff our SNH, patients are able to access to similar quality of care. In addition, multidisciplinary neuro-oncology clinics and tumor boards at our SNH allow for coordination of patient care while applying evidence-based guidelines. Other SNHs may benefit from adopting a multidisciplinary neuro-oncology care model in order to increase rates of standard of care treatment of GB.

Interestingly, there was a small subset of long-term survivors at the SNH that was not seen at the PH. Stupp et al demonstrated 5 year OS of 10% with combined chemoradiotherapy with temazolamide.¹ This is similar to the 5 year OS in patients treated at our SNH; there were fewer long-term survivors at our PH. One contributing factor could be difference in the age in patients. Among clinical factors affecting prognosis, older age and poor performance status has consistently been shown to be associated with shorter survival.^3,24 Thirty of 39 patients (77%) were older than 50 years of age at the PH compared to 36 of 55 (65%) at the SNH. On the other hand, there was no difference in KPS at presentation between each patient cohort. This is reassuring since a similar study had demonstrated that patients at SNH presented with poorer KPS, which negatively impacted their survival.⁴ Though our study did not find that age greater than 50 years was not statistically significant prognostic factor for survival in multivariate setting, this has been validated as a poor prognostic factor in other studies. We hypothesize that this is because our sample size was not large enough to detect a significant difference.

A limitation to our study that might explain the difference in long-term survival is that molecular analysis was not routinely performed on the tumors at the PH (16 patients) and was not performed at all at the SNH. MGMT is associated with improved survival compared with unmethylated MGMT.²⁵ In addition, patients with IDH mutation have a better prognosis. The potential for differences in MGMT methylation may explain between the two patient cohorts. However, the likelihood of this occurring is less likely despite the racial differences given MGMT methylation was not different between Latinos and non-Latino in two institutional studies.^15,16 Given the heterogeneity of our cohorts, there may be undetermined differences in the molecular pathways of GB in non-White patients, which may explain similar oncologic outcomes despite lower rates of adjuvant therapy at our SNH. Future genomic investigation to elucidate this hypothesis is warranted.

Other limitations to this study are the fact it was retrospective and a relatively small sample size. SNH and PH patients had significantly different baseline characteristics for which we attempted to account for in the multivariate analysis. Therefore, the results may or may not be generalizable to other practice settings where indigent patients are managed, potentially compromising the external validity of our study. Further study in this area in other indigent care settings is needed to confirm our findings. Lastly, there is ascertainment bias due pattern of care. Many patients that were excluded from our study may have chosen to get treatment closer to their primary residence given the need for daily radiation, where as patients who presented emergently for surgery at our SNH may be required to receive their adjuvant therapy at an in-network provider. Hispanic residents may return to their native countries for therapy or other reasons may potentially skew the composition of the patient cohort.

In this retrospective cohort of GB patients, we observed obvious SES disparities between patients and differences in the rate of delivery in adjuvant therapies. In order to improve compliance with adjuvant therapy, our SNH system has implemented a dedicated Patient Navigator Program in 2013 for patients to ensure that these medically underserved cancer patients and their families are connected with essential information, resources and support that helps to reduce barriers to care. Nevertheless, despite these disparities in care, this did not lead to significantly different survival outcomes. We believe that this is because, at least in part, our SNH patients with GB are under the centralized care at an academic center with the vast majority of patients discussed at a weekly multidisciplinary brain tumor board and followed by the multidisciplinary neuro-oncology service. Similar neurosurgical care was provided at both the SNH and PH sites, including the same faculty neurosurgeons, identical equipment and capability for speech and motor mapping with awake craniotomy as needed, though 5-ALA and intraoperative MRI was not utilized at each site. By being seen in a multidisciplinary neuro-oncology service, SNH patients are evaluated for adjuvant treatment at their initial post-operative visit, minimizing the need for multiple visits as well minimizing delays in starting treatment by optimizing care coordination and the referral process. Review at our multidisciplinary brain tumor board allows for further care coordination between care teams and consideration for clinical trials. Improving access to this care in other underserved communities would hopefully bridge the disparities in care in disadvantaged patients. Future directions include following up on the SNH patients with long-term survival and performing molecular analysis on these tumors in order to understand the pathogenesis and prognosis of GB in this population.

Highlights.

• GB patients at our SNH received less adjuvant treatment compared to PH.

• Rates of adjuvant chemoradiation at our SNH are higher than other SNHs in other studies.

• Better compliance to standard of care treatment may explain similar PS and OS between our SNH and PH.

Glossary

SES

Socioeconomic Status

GB

Glioblastoma

SNH

Safety-Net Hospital

PH

Private Hospital

OS

Overall Survival

PFS

Progression-Free Survival

TMZ

Temozolomide

MGMT

O6-methylguanine DNA methyltransferase

KPS

Karnofsky Performance Scores

RPA

Recursive Partioning Analysis