Association of immunoglobulin E levels with glioma risk and survival

Abstract

Background: Previous epidemiologic studies have reported an association of serum immunoglobulin E (IgE) levels with reduced glioma risk, but the association between IgE and glioma prognosis has not been characterized. This study aimed to examine how sex, tumor subtype, and IgE class modulate the association of serum IgE levels with glioma risk and survival.

Methods: We conducted a case-control study using participants from the University of California, San Francisco Adult Glioma Study (1997-2010). Serum IgE levels for total, respiratory and food allergy were measured in adults diagnosed with glioma (n=1319) and cancer-free controls (n=1139) matched based on age, sex, and race and ethnicity. Logistic regression was adjusted for patient demographics to assess the association between IgE levels and glioma risk. Multivariable Cox regression adjusted for patient-specific and tumor-specific factors compared survival between the elevated and normal IgE groups. All statistical tests were 2-sided.

Results: Elevated total IgE was associated with reduced risk of IDH-wildtype (RR=0.78, 95% CI: 0.71-0.86) and IDH-mutant glioma (RR=0.73, 95% CI: 0.63-0.85). In multivariable Cox regression, positive respiratory IgE was associated with improved survival for IDH-wildtype glioma (RR=0.79, 95% CI: 0.67-0.93). The reduction in mortality risk was significant in females only (RR=0.75, 95% CI: 0.57-0.98) with an improvement in median survival of 6.9 months (P<.001).

Conclusion: Elevated serum IgE was associated with improved prognosis for IDH-wildtype glioma, with a more pronounced protective effect in females than males, which has implications for the future study of IgE-based immunotherapies for glioma.

Figure 1:. Associations between IgE levels and survival stratified by sex and molecular subtype.

Risk ratios (RR) and corresponding 95% confidence intervals (CI) were estimated using IPTW Cox regression models. Propensity models include age, sex (included in combined analysis only), self-reported race and ethnicity, and chemotherapy use. Weighted Cox models were also adjusted for dexamethasone use, surgery type (resection/biopsy only), 1p19q codeletion status, TERT mutation, tumor grade and AGS recruitment series. The RR estimate for food IgE and IDH-mutant glioma (combined and female only) are not shown due to insufficient individuals with positive food IgE.

Figure 2:. Kaplan-Meier curves for adult glioma cases with serum IgE concentrations.

Percent survival distributions for adults with glioma stratified by sex and categorical serum IgE concentration for (A) IDH-wildtype glioma and total IgE and (B) IDH-wildtype glioma and respiratory IgE. +IgE = above normal total IgE or positive respiratory IgE. −IgE = normal total IgE or negative respiratory IgE. P values are shown for the sex-stratified Kaplan-Meier curves.

Figure 3:. Population attributable fractions (PAF) for IDH-WT survival stratified by sex and follow-up time.

PAF estimates for age at diagnosis (<58 years/≥58 years), surgery (resection/biopsy), chemotherapy use (yes/no), and respiratory IgE (positive/negative) were derived from Cox proportional hazard regression models adjusted for sex (combined only), self-reported race and ethnicity, radiation use, 1p19q codeletion status, TERT mutation status, and strata for tumor grade. PAF estimates are reported for 6 months and 12 months of follow-up time. Estimates that do not achieve P<0.05 are indicated by a striped pattern.