Baseline characteristics of the GBM patients

A total of 6190 patients diagnosed with GBM (IDH-wildtype) between 2018 and 2021 were assessed from the SEER database. The overall characteristics of these patients were summarized in Table 1. There were 1947 (31.4%) patients in the low-income group, 2413 (39%) patients in the middle-income group, and 1830 (29.6%) in the high-income group. We found that race, marital status, overall survival, and GBM-specific death were significantly different among the three groups (p < 0.001, p = 0.039, p = 0.040, and p = 0.003, respectively, Supplementary Table 1). The low-income group had a higher proportion of Black individuals, the middle-income group had the highest proportion of unmarried individuals, the high-income group exhibited the highest average survival, and the low-income group had the highest rate of GBM-specific death.

Table 1 Baseline characteristics of GBM patientsUnivariate and multivariate Cox regression on overall survival

Univariate analyses showed that increased age and male sex were significantly associated with worse overall survival, while married status and higher household income were significantly associated with better overall survival (Table 2, Supplementary Fig. 1A-D). Then we included the significant variables into the multivariate regression model. Multivariate analyses identified age, male sex, marital status, and household income as independent factors for overall survival. Compared to the low-income group, the middle-income group and high-income group showed significantly better survival (HR 0.881 [0.816–0.951], p = 0.001; HR 0.860 [0.792–0.934], p < 0.001; Fig. 2A; Table 2). The nomogram shows how age, sex, marital status, and household income contribute to a patient’s survival (Supplementary Fig. 2).

Table 2 Univariate and multivariate Cox regression analyses for overall survivalFig. 2

Forest plots for overall survival and GBM-specific survival. A-B. The forest plot shows that old age, male sex, unmarried status, and lower-income household contributed to worse overall survival (A) and GBM-specific survival (B) for patients with IDH-wildtype GBM

Univariate and multivariate Cox regression on GBM-specific survival

Tumor-specific death remained the predominant cause of mortality with worse survival. We further investigated whether these factors affect GBM-specific death in the cohort. The univariate analyses exhibited that increased age and male sex significantly increased the risk of GBM-specific survival; however, married status and higher household income significantly decreased the risk of GBM-specific survival (Table 3, Supplementary Fig. 1E-H). In multivariate analyses, old age, male sex, and unmarried status were significantly independent risk factors for GBM-specific survival. Furthermore, the middle- and high-income groups exhibited better GBM-specific survival compared to the low-income group (HR 0.866 [0.799–0.937], p < 0.001; HR 0.854 [0.784–0.929], p < 0.001; Fig. 2B; Table 3).

Table 3 Univariate and multivariate Cox regression analyses for GBM-specific survivalSubgroup analyses among different years

As advancements continue in the diagnosis and treatment of GBM, we delved into how household income impacts GBM survival over time. The proportions of the three income groups remained relatively stable from 2018 to 2021, and the age, sex, and marital status distribution of patients did not show significant changes. However, there was an increasing trend in the proportion of patients aged over 60 years, along with a decreasing trend in both overall survival and GBM-specific death (Supplementary Table 2). Our analysis revealed that, when compared with the low-income group, the HRs (all < 1) for overall survival and GBM-specific death in the middle- and high-income groups showed a decreasing trend over the years, indicating an increasing disparity across household income groups. The risk for overall survival in the middle-income group decreased below the level of significance after 2020, which is one year later than in the high-income group (after 2019). However, the HRs for both groups decreased below the level of significance for GBM-specific survival since 2019 (Supplementary Fig. 3A-B).

Interaction effect between household income and other variables

To explore the interaction effect, household income and age were reclassified as binary variables based on the cutoffs of $80,000 and 60 years old. Patients were stratified into four groups: age < 60 & household income ≥ $80,000, age < 60 & household income < $80,000, age ≥ 60 & household income ≥ $80,000, and age ≥ 60 & household income < $80,000. The Kaplan-Meier survival curves based on the Cox model demonstrated a more obvious overall survival and GBM-specific survival difference across age groups than household income groups. Compared to the age < 60 & household income ≥ $80,000 subgroup, the other three subgroups exhibited worse overall survival (OS: HR 1.06 [0.94, 1.20], p = 0.3426; HR 1.73 [1.59, 1.88], p < 0.001; HR 2.06 [1.87, 2.26], p < 0.001) and GBM-specific survival (HR 1.06 [0.93, 1.20], p = 0.3828; HR 1.71 [1.57, 1.87], p < 0.001; HR 2.07 [1.88, 2.29], p < 0.001, Fig. 3A-B). The effect of household income on survival was observed in the age < 60 subgroup (OS: HR 1.19 [1.09, 1.29], p < 0.001; GBM-specific survival: HR 1.21 [1.11, 1.32], p < 0.0001). While the effect of age on survival generally existed within strata of household income (OS: HR 1.73 [1.59, 1.88], p < 0.001; HR 1.94 [1.72, 2.19], p < 0.0001; GBM-specific survival: HR 1.71 [1.57, 1.87], p < 0.001; HR 1.96 [1.73, 2.22], p < 0.0001; Tables 4 and 5).

Fig. 3

Subgroup survival analyses for different age and household income groups. A-B. Kaplan-Meier Survival curves show an obvious survival difference between age groups compared to household income groups (A. overall survival, B. GBM-specific survival, p < 0.0001, respectively). When age ≥ 60, patients with lower-income households exhibited much worse overall survival (A) and GBM-specific survival (B)

Table 4 Interaction of age and household income in multivariate Cox model for overall survivalTable 5 Interaction of age and household income in multivariate Cox model for GBM-specific survival

We found evidence of a synergistic additive interaction between household income and age on OS and GBM-specific survival. Compared with the referenced group, the RERI was 0.26 ([0.06, 0.46], p = 0.0051) and 0.31 ([0.10, 0.51], p = 0.0019), indicating that the interaction contributed to 26% of the total effect on OS and 31% on GBM-specific survival. The AP was 0.13 ([0.03, 0.22], p = 0.0040) for OS and 0.15 ([0.05, 0.24], p = 0.0013) for GBM-specific survival. The corresponding SI was 1.33 ([1.05, 1.70], p = 0.0099) and 1.40 ([1.08, 1.81], p = 0.0050), suggesting that the joint effect exceeded the sum of their independent effects by 33% and 40%. Multiplicative scales for the interaction between household income and age were not significant for OS and GBM-specific survival (OR 1.15, 95% CI [0.98, 1.34], p = 0.1340; OR 1.15, 95% CI [0.98, 1.34], p = 0.0777, respectively, Tables 4 and 5).

Furthermore, household income showed a significant effect on OS and GBM-specific survival within the strata of sex and marital status; however, no multiplicative or additive interaction effect was identified in the interaction analyses (Supplementary Tables 3–6).