Immune phenotypes predict survival in patients with glioblastoma multiforme

Abstract

Background: Glioblastoma multiforme (GBM), a common primary malignant brain tumor, rarely disseminates beyond the central nervous system and has a very bad prognosis. The current study aimed at the analysis of immunological control in individual patients with GBM.

Methods: Immune phenotypes and plasma biomarkers of GBM patients were determined at the time of diagnosis using flow cytometry and ELISA, respectively.

Results: Using descriptive statistics, we found that immune anomalies were distinct in individual patients. Defined marker profiles proved highly relevant for survival. A remarkable relation between activated NK cells and improved survival in GBM patients was in contrast to increased CD39 and IL-10 in patients with a detrimental course and very short survival. Recursive partitioning analysis (RPA) and Cox proportional hazards models substantiated the relevance of absolute numbers of CD8 cells and low numbers of CD39 cells for better survival.

Conclusions: Defined alterations of the immune system may guide the course of disease in patients with GBM and may be prognostically valuable for longitudinal studies or can be applied for immune intervention.

Keywords: CD39-ectonucleotidase; CD8+ lymphocytes; Glioblastoma multiforme; NK cells; Recursive partitioning analysis; Survival.

Fig. 1

Increased median value of IL-10 and ferritin in plasma samples of GBM patients. Interleukin 10 (IL-10) (a), plasma ferritin (b) as relevant biomarkers determined in plasma samples of GBM patients (P) and healthy controls (H). Median values for IL-10 is 7.18 (±1.2 pg/ml) in elderly healthy donors, and ferritin serum concentrations may range a lot in elderly. IDH-1 mutated patients outside of the 25–75 boxed percentile are shown as red-filled circles

Fig. 2

Significantly decreased CD3 and TCR α/β in peripheral blood cells but no significant change in CD95. Percent CD3 lymphocytes in EDTA whole-blood-derived lymphocytes of GBM patients (P) and controls (H) (a, left) and the corresponding CD3 expression densities (mean fluorescence intensity (MFI)) (a, middle) and absolute numbers of CD3-positive lymphocytes in patients (a, right). Normal ranges of absolute numbers are indicated as blue arrowheads on the y-axis (a, right). Percent TCRα/β lymphocytes in GBM patients and healthy controls (b, left) and the corresponding MFI distribution (b, middle) and absolute numbers of TCRα/β cells/μl blood (b, right). Percent CD95 lymphocytes in GBM patients as compared to controls (c, left) and the corresponding distribution of CD95 MFI (c, middle) and absolute CD95-expressing lymphocytes (c, right). IDH-1 mutated patients outside of the 25–75 boxed percentile are shown as red-filled circles. IDH-1 mutated patients outside of the 25–75 boxed percentile are shown as red-filled circles

Fig. 3

Differences in CD56 and CD16/56 NK cells and CD3/56 cytokine-activated NK (CIK) cells in GBM patients and controls. Percent CD cluster-positive lymphocytes are shown for patients (P) and healthy donors (H) on the left; relative expression densities, given as MFI are shown in the middle column, and absolute numbers of CD cluster-positive lymphocytes for the patients are shown on the right half of the figure. CD56 NK cells (a). CD56/16 co-expressing, activated NK cells (b). CD3/56 co-expressing, cytokine-induced killer (CIK) cells (c). IDH-1mutated patients outside of the 25-75 boxed percentile are shown as red-filled circles

Fig. 4

Differences in CD8, CD4, CD4/25, CD39, and CD127-positive lymphocytes. Percent CD cluster-positive lymphocytes are shown for patients (P) and healthy donors (H) on the left; relative expression densities, given as MFI are shown in the middle column, and absolute numbers of CD cluster positive lymphocytes for the patients are shown on the right half of the figure. CD8 (a). CD4 (b). CD4+/CD25+ (c). CD39 (d). CD127 (e). IDH-1 mutated patients outside of the 25-75 boxed percentile are shown as red-filled circles

Fig. 5

Kaplan-Meier curves of overall survival and lymphocyte subpopulations, IDH-1 and MGMT mutation status. Overall survival of 51 patients with GBM (a), distinguished by relative amounts of CD expressing lymphocytes. Patients were grouped according to the median values of markers determined by Mann-Whitney U statistical analysis. CD8 cells (b), TCRα/β T cells (c), CD95 expressing lymphocytes (d), CD16/56 activated NK cells (e), and CD127 expressing lymphocytes (f); IDH-1 mutation status (g) and MGMT mutation status (h)

Fig. 6

Schematic representation of effector cells guiding immune surveillance in GBM and counter regulatory CD39-positive cells. A number of potential effector cells have been studied for their differential expression in GBM patients at the time of diagnosis. Based on regression analysis of survival data based on the Cox proportional hazards model and model-based recursive partitioning, the beneficial effects of high numbers of CD8-positive lymphocytes and the negative effect by CD39-positive lymphocytes has been demonstrated. Anti-tumor reactive CD8 cells may migrate between the brain and the peripheral tissues. They are likely to be counter regulated by regulatory lymphocytes expressing CD39. CD39 cells may also migrate between tissues or may be induced by immune suppressive biomarkers which can be cytokines but also extracellular vesicles. Tumor tissue is a likely source of such biomarkers.