Gefitinib selectively inhibits tumor cell migration in EGFR-amplified human glioblastoma

Abstract

Background: Tissue invasion is a hallmark of most human cancers and remains a major source of treatment failure in patients with glioblastoma (GBM). Although EGFR amplification has been previously associated with more invasive tumor behavior, existing experimental models have not supported quantitative evaluation of interpatient differences in tumor cell migration or testing of patient-specific responses to therapies targeting invasion. To explore these questions, we optimized an ex vivo organotypic slice culture system allowing for labeling and tracking of tumor cells in human GBM slice cultures.

Methods: With use of time-lapse confocal microscopy of retrovirally labeled tumor cells in slices, baseline differences in migration speed and efficiency were determined and correlated with EGFR amplification in a cohort of patients with GBM. Slices were treated with gefitinib to evaluate anti-invasive effects associated with targeting EGFR.

Results: Migration analysis identified significant patient-to-patient variation at baseline. EGFR amplification was correlated with increased migration speed and efficiency compared with nonamplified tumors. Critically, gefitinib resulted in a selective and significant reduction of tumor cell migration in EGFR-amplified tumors.

Conclusions: These data provide the first identification of patient-to-patient variation in tumor cell migration in living human tumor tissue. We found that EGFR-amplified GBM are inherently more efficient in their migration and can be effectively targeted by gefitinib treatment. These data suggest that stratified clinical trails are needed to evaluate gefitinib as an anti-invasive adjuvant for patients with EGFR-amplified GBM. In addition, these results provide proof of principle that primary slice cultures may be useful for patient-specific screening of agents designed to inhibit tumor invasion.

Keywords: EGFR; gefitinib; glioblastoma; invasion; migration; personalized therapy; slice culture.

Fig. 1.

Time-lapse confocal microscopy of tumor cells in human glioblastoma (GBM) slice cultures allows for tracking of cell migration in the tumor microenvironment. (A) Tumor cells expressing ZsGreen in human GBM slice cultures were visualized using confocal laser microscopy at 72 h after retroviral infection. A representative 10× microscope field is shown. (B) A migration path map was generated by tracking all tumor cells observed using the time-lapse images recorded from the tumor microregion in panel A. (C) The motility index, or percentage of cells that moved at least a cell width (10 μm) during the imaging, was calculated for each patient tumor cell population tracked (n = 14). (D) A migration displacement map was generated by centering all tumor cell migration paths at a common starting point (scale is indicated on each axis).

Fig. 2.

Heterogeneity exists in tumor cell migration speed, directionality, and effective migration speed among patients with glioblastoma (GBM). (A) Cell migration speed varied significantly across our cohort of tracked tumor cell populations (n = 14, ANOVA P < .0001). Bars represent the speed of the 10th to 90th percentile cells, the line represents the median, and the cross represents the mean cell migration speed of the tumor cell population. (B) The percentage of cells migrating faster than 5 μm/h varied among tumor cell populations. Tumor cell populations from patients with GBM are ordered left to right according to increasing mean cell migration speed. (C) In a subset of tumor cell population, there was an increased number of cells that migrated faster than 10 μm/h. (D) Graphical representation of the migration directionality, depicting cells with less efficient (top, directionality <1) and more efficient (bottom, directionality = 1) patterns of migration. (E) Directionality of tumor cell migration within organotypic slices varied significantly among patients (n = 14, ANOVA P < .0001). (F) Effective migration speed varied ∼7-fold across our patient cohort (n = 14, ANOVA P < .0001).

Fig. 3.

EGFR amplification is associated with augmented migration in human glioblastoma (GBM), and selective targeting of EGFR reduces tumor cell migration in receptor-amplified patients. Comparison of mean migration speed (A), directionality (B), and effective migration speed (C) among EGFR-amplified and nonamplified tumors. Directionality and effective migration speed were significantly increased in amplified tumors (P < .05 for both metrics), whereas migration speed trended toward an increase in amplified tumors (P = .14). (D) Representative displacement maps generated during imaging of pretreated (DMSO) and posttreated (gefitinib) conditions, demonstrating differential qualitative effects of EGFR inhibition on tumor cell migration in slices from EGFR-amplified (D) and nonamplified (E) GBM.

Fig. 4.

Gefitinib exhibits selective effects on tumor cell migration in EGFR-amplified glioblastoma (GBM). Quantitative analysis of the effects of gefitinib on mean migration speed (A), directionality (B), and effective migration speed (C) for EGFR-amplified (red circles) and nonamplified (black squares) tumors. Mean migration speed and effective migration speed significantly decreased in each EGFR-amplified tumor with treatment (P < .02 for all tumors), whereas no decrease was observed in nonamplified tumors. One EGFR-nonamplified tumor demonstrated a significant change in directionality (P = .005). Error bars represent standard error of the mean (SEM). Representative analysis of tumor cells in nonamplified (D) and amplified (E) tumors before (DMSO) and after gefitinib (Gef) treatment, subdivided by migration speed into 5 μm/h bins, confirming selective dropout of faster migrating cells after treatment in EGFR-amplified tumors. Analysis of each of these tumor cell populations before and after gefitinib treatment via the Kolmogorov-Smirnov test demonstrated the EGFR-amplified cell population had a significant (P < .0001) shift in the distribution of cell population speeds, whereas the EGFR-nonamplified population demonstrated no shift (P = .85).