Ex vivo drug sensitivity screening predicts response to temozolomide in glioblastoma patients and identifies candidate biomarkers

Abstract

Background: Patient-derived glioma stem-like cells (GSCs) have become the gold-standard in neuro-oncological research; however, it remains to be established whether loss of in situ microenvironment affects the clinically-predictive value of this model. We implemented a GSC monolayer system to investigate in situ-in vitro molecular correspondence and the relationship between in vitro and patient response to temozolomide (TMZ).

Methods: DNA/RNA-sequencing was performed on 56 glioblastoma tissues and 19 derived GSC cultures. Sensitivity to TMZ was screened across 66 GSC cultures. Viability readouts were related to clinical parameters of corresponding patients and whole-transcriptome data.

Results: Tumour DNA and RNA sequences revealed strong similarity to corresponding GSCs despite loss of neuronal and immune interactions. In vitro TMZ screening yielded three response categories which significantly correlated with patient survival, therewith providing more specific prediction than the binary MGMT marker. Transcriptome analysis identified 121 genes related to TMZ sensitivity of which 21were validated in external datasets.

Conclusion: GSCs retain patient-unique hallmark gene expressions despite loss of their natural environment. Drug screening using GSCs predicted patient response to TMZ more specifically than MGMT status, while transcriptome analysis identified potential biomarkers for this response. GSC drug screening therefore provides a tool to improve drug development and precision medicine for glioblastoma.

Fig. 1. Correlation of molecular features of parental tissues and derived cell cultures.

a Correlation analysis of DNA sequences between the tumours and their corresponding cell cultures based on the normalized logR ratios (n = 19) (mean Pearson’s r = 0.77). The values represent the correlation coefficients. Samples names with “C” indicate cell cultures, while “T” indicate tissues. b Correlation analysis of the log2 averaged gene expression between tumours and corresponding cell cultures (n = 19) (Pearson’s r = 0.76, p < 0.0001). The blue line represents the trendline of the linear regression. c Gene Set Enrichment Analysis (GSEA) clustermap illustrating differential expression of bulk RNAseq data between corresponding tissue (n = 19) and cell cultures (n = 19) in reference to the MSigDB_Hallmark_2020 library extracted from Enrichr. Genes from the top 4 terms containing the most negatively and positively normalized enrichment scores (NES) are represented here (n = 755). d GSEA illustrates the top 20 terms with the most positively (n = 10) and negatively (n = 10) NES. Blue indicates terms with negative enrichment scores and red positive enrichment scores. Intensity of the colour is indicative of relative significance (all terms present are NOM p-val, FDR q-val, FWER p val significant). e Violin plots illustrating mean expression of hallmark genes between tissue and cell cultures. Each line indicates an individual paired sample. P values < 0.05 were considered significant.

Fig. 2. MGMT status of primary tumours is generally maintained in derived GBM cultures and related to in vitro response to TMZ.

a Percentage of GSC cultures (n = 64) with maintained MGMT promoter methylation status (Retained), switched from methylated to unmethylated or vice versa (Switched) or selection from mixed to single signal (Selection). b In vitro response of TMZ (expressed as % cell viability at 100 uM, y axis) compared to the MGMT status of GSC cultures. c In vitro response of TMZ (expressed as AUC, y axis) compared to the MGMT status of GSC cultures. d In vitro response of TMZ (expressed as IC₅₀, y axis) compared to the MGMT status of GSC cultures. Asterisks indicate significantly different means at the indicated read-out (unpaired t test, ***p = 0.0006, **p = 0.002 and *p = 0.02). P values < 0.05 were considered significant.

Fig. 3. In vitro response to TMZ predicts clinical response of corresponding patients.

Correlation analysis of the percentage cell viability at 100 uM TMZ (y-axis, percentage of DMSO control) of GSC cultures (n = 55) with (a) progression free survival (x-axis, PFS in months) or (b) the overall survival (x-axis, OS in months) of the corresponding patients (Spearman’s rank correlation). c Survival graph of patient cohort (n = 55), comparing overall survival (x axis, OS in months) to TMZ in vitro response as classified by responders (n = 9, green), intermediates (n = 18, blue) and non-responders (n = 28, red), Responders vs intermediate (log-rank p = 0.0023), intermediate vs non-responders (log-rank p = 0.0082) and responders vs non-responders (log-rank p < 0.0001). d Distribution of MGMT status of GSC cultures of the three response groups (responders, intermediates and non-responders). e Survival graph of patient cohort (n = 55) comparing survival (x-axis, OS in months) to cell culture MGMT status, unmethylated (red) vs methylated (blue) (log-rank p = 0.0019). f Correlation of MGMT RNA expression level to % cell viability at 100 μM of TMZ (R² = 0.43, p = 0.0024). g Comparison of MGMT RNA expression level to the MGMT promoter status (M and UM) (p = 0.0106). P values < 0.05 were considered significant.

Fig. 4. In vitro sensitivity of TMZ predicts treatment response in patients who received TMZ/RTx.

Representative examples of consecutive T1 weighted post gadolinium MRI images of 6 patients, which were defined as 3 predicted responders (cell viability below 50%, green curves, left graph) and 3 non-responders (cell viability above 75%, red curves, right graph), showing stable disease in the responder patients (left) and progressive disease in the non-responder patients (right). Tumour area was delineated in all MRI scans, responders (green) and non-responders (red).

Fig. 5. Transcriptome correlations between in vitro and patient response to TMZ.

a Heatmap of the top 100 signature genes correlated with TMZ response in terms of the % viability at 100 uM for the 19 GBM cell cultures (p value < 0.05). Each row of the heatmap represents one significant gene, while each column represents one sample. The genes are ordered by the coefficients of the Spearman correlation analysis. The samples are ordered by the values of the TMZ response outcomes. b Venn diagram describes the significant correlated genes to TMZ response in both tissue and cell culture datasets as well as the overlapping genes. c Overlapping pathways in cell cultures (left) and the tumours (right) (False discovery rate <0.25). The FDR p values on the x-axis show the significance of the pathways in cell cultures/tumours, while the size of the circles represents the enrichments score of each pathway in each category. The pathways are ordered by the averaged enrichment scores of the cell cultures and tumours.