Dynamic expression of 11 miRNAs in 83 consecutive primary and corresponding recurrent glioblastoma: correlation to treatment, time to recurrence, overall survival and MGMT methylation status

Our retrospective study included eighty-three patients with recurrent GBM. The data were obtained from the patients charts from hospital registry of the Institute of Oncology and from the Cancer Registry of Slovenia. Each patient underwent the first surgery and the second one after the first recurrence of GBM between January 1997 and November 2011 at the University Clinical Centre Ljubljana, Slovenia. The selected patients were consecutive; of the 1117 that underwent the first surgery for GBM, 83 underwent also the second surgeryat tumor recurrence since these patients were capable of re-operation according to clinical standards (Karnofsky performance status over 80). Patients received radiotherapy (RT) after the surgery, and RT with concomitant and adjuvant chemotherapy (ChT) after 2005.

Tissue samples were fixed in 10% buffered formalin and embedded in paraffin (FFPE) for routine diagnostics. The grading criteria based on WHO classification was used.14 Paired tissue samples were retrieved from the archive of the Institute of Pathology, Faculty of Medicine, University of Ljubljana. Samples were classified in two groups, i.e. primary GBM (sample tissue collected at first surgery) and corresponding recurrent GBM (sample tissue collected at second surgery). As control sample RNA FirstChoice Human Brain Reference Total RNA (Cat. no. 6050, Ambion; Invitrogen, USA) was used.

The patients’ data, including age, gender and therapy are summarized in Table 2.

Tissue samples were cut at 10 μm from FFPE tissue blocks and for the isolation procedure six to eight 10 μm sections were used. Total RNA isolation was performed using miRNeasy FFPE Kit (Qiagen, Germany) according to the manufacturer’s protocol. The RNA was eluted in 30 μl of nuclease-free water. The yield was measured spectrophotometrically using the NanoDrop-1000 (Thermo Scientific, USA) and the quality was evaluated on the Bioanalyzer 2100 (Agilent Technologies, USA).

All the reagents were from Qiagen, except where otherwise indicated. Quantitative PCR (qPCR) was carried out using the Rotor Gene-Q Real-Time PCR System and all the qPCR reactions were performed in duplicates or triplicates. Prior to qPCR analysis, two pools of RNA samples were created from FFPE tissue samples (primary, recurrence) and, along with reference RNA, tested for qPCR efficiency. Reverse transcription using miScript reverse transcription kit was performed in a 10 μl reaction master mix with 50 ng of total RNA according to manufacturer instruction. The resulting cDNA was diluted 100-fold and qPCR reaction was carried out in 10 μl PCR master mix, according to manufacturer instruction. As the reference genes, RNU6B and SNORD25 were used based on efficiency results. Tested miRNAs were: miR-7, miR-9, miR-15b, miR-21, miR-26b, miR-124a, miR-199a, let-7a, let-7b, let-7d, and let-7f.

In a subset of patients that were treated with temozolamide (TMZ) (n = 47), MGMT methylation status was determined. Tissue samples from the first surgery and the surgery at first recurrence were cut at 10 μm from FFPE tissue blocks and for the isolation procedure, six to eight 10 μm sections were used. Total DNA isolation was performed using QIAamp DNA FFPE Tissue Kit (Qiagen) according to the manufacturer’s protocol. The DNA was eluted in 60 μl of nuclease-free water. The yield was measured fluorescently using the Quant-It (Life Technologies) according to manufacturer instruction and Rotor Gene Q (Qiagen).

For determining the MGMT methylation status, methyl-specific polymerase chain reaction (MSP) was used in a two-step approach with primers previously described.16 Briefly, prior to MSP, 500 ng of DNA was used for bisulfite conversion using innuCONVERT Bisulfite Basic Kit according to manufacturer instruction (Analytik Jena) and stored at -20°C for subsequent MSP. For MSP, 15 ng of bisulfite converted DNA was used with 0.2 μM of each primer for methylated form and 0.3 μM of each primer for unmethylated form, 2 mM of dNTP and 0.25 U of Hot Master Polymerase (5 Prime), all in 10 μl reaction. Amplification was performed according to manufacturer instruction using 59°C for primer annealing. In each run, fully methylated (EpiTect Control DNA, methylated, Qiagen) as well as fully unmethylated controls (EpiTect Control DNA, unmethlyated, Qiagen) were used as assay controls. Results were analyzed using 2% agarose gel. The investigator who analyzed the GBM samples was blinded to all clinical information.

To present a relative gene expression the 2^-ΔΔCt method was used.17 For paired samples (primary, recurrent), the calculated ΔCt was tested for statistical significance using Wilcoxon Signed Rank test. For independent group of samples, ΔΔCt was calculated relatively to Human brain Reference RNA and used for Mann-Whitney test (treatment, MGMT status, etc.). Kaplan-Meier curve was used to analyze difference in progression free survival (time to recurrence) and overall survival (different treatment options, miRNA expression). Correlations between scale variables (miRNA expression, survival) were calculated using Pearson’s and Spearman’s correlation coefficient. For all statistical tests, the SPSS analytical software (ver.24 SPSS Inc. IL) was used (cut-off point at p < 0.05). The workflow of comparisons is summarized in Figure 1.

FIGURE 1

Study has been approved by Slovene National Medical Ethics Committee (Number 113/05/13).

Comparing expression of analyzed miRNAs to Human Brain Reference RNA (ΔΔCt), we found altered expression for all 11 analyzed miRNAs both in primary as well as in recurrent GBM. Whereas miR-9 and miR-21 were mainly up-regulated and miR-15b was mainly down-regulated in primary and recurrent GBM, all other miRNAs showed more heterogeneous expression when compared to the Human Brain Reference RNA. Results are summarized in Figure 2.

FIGURE 2

Using calculated ΔCt and Wilcoxon Signed Rank test, we have observed significant alteration in expression for 7 miRNAs between primary and recurrent GBM. Five miRNAs, miR-9 (p = 0.016), miR-21 (p < 0.001), miR-26b (p < 0.001), miR-124 (p = 0.029) and miR-199a (p < 0.001), were down-regulated, and miR-7 (p = 0.001) and let-7f (p < 0.001) were up-regulated in recurrent tumors compared to first time surgery. However, all of these miRNAs were differentially expressed also in a subgroup treated with RT and ChT, whereas only four of these showed statistically significant differential expression in a subgroup treated with RT alone. Results are summarized in Figure 3 and Figure 2 (Average Fold Change Table).

FIGURE 3

Using Mann-Whitney test for comparing subgroup of patients treated with RT alone to patients treated with RT and ChT (analyzed miRNAs normalized to Human Brain Reference RNA, ΔΔCt), we observed that expression of certain miRNAs was different before treatment (at first surgery) as well as after treatment (at second surgery). Therefore, only those miRNAs that showed significantly differential expression only after treatment (at second surgery) were considered truly as a consequence of treatment. Both miR-15b and miR-7d were down-regulated compared to Human Brain Reference RNA after treatment, and there was significant change in expression between patients that received RT alone and those that received RT and ChT (p = 0.013 for miR-15b and p = 0.008 for let-7d). In patients treated with RT alone, let-7f was up-regulated compared to Human Brain Reference RNA at second surgery, whereas it was unchanged in patients that were treated with RT and ChT. There was significant change in expression of let-7f between both groups (p = 0.007). Results are summarized in Figure 4.

FIGURE 4

There was weak, but statistically significant change in progression free survival between patients who received RT alone and those who received both, RT and ChT after first surgery (7 months and 11 months, respectively; p = 0.045). Results are summarized in Figure 5A

FIGURE 5

We have also calculated the survival time from second surgery. The number of patients who received RT alone or RT and ChT was too small for reliable analysis of survival time. However, there was statistically significant longer overall survival for patients who received ChT compared to those that were not treated after second surgery (8 months and 5 months, respectively; p = 0.004). Results are summarized in Figure 5B.

We further calculate Spearman’s and Pearson’s correlation coefficient between miRNA expression (analyzed miRNAs normalized to Human Brain Reference RNA, ΔΔCt) in primary GBM and progression free survival and overall survival, and found weak inverse correlation for let-7b and progression free survival (r = -0.220, p = 0.049) and weak inverse correlation between let-7f and overall survival (r = -0.226, p = 0.050). We further analysed difference in Kaplan-Meier curve for patients with up- and down-regulated miRNAs and found, that there is a difference in progression free between patients with up- and down-regulated miR-26b (p = 0.05, Figure 5C), and difference in overall survival for up- and down regulated let-7f (p = 0.021).

Among patients treated with ChT, there were 47 treated with TMZ. Among them, there were 9 patients with unmethylated and 34 with methylated MGMT. All, except one, became methylated in recurrent tumor. Comparing miRNA expression normalized to Human Brain Reference RNA between MGMT methylated and MGMT unmethylated tumors revealed no correlation of investigated miRNA expression to MGMT methylation status.

We observed differential expression of the majority of analyzed miRNAs in both samples. In primary as well as in recurrent GBM we have observed overall miR-7 over-expression, although was expression heterogeneous among tumors. In spite of numerous studies on miR-7 expression, its expression in recurrent GBM has not been analyzed yet. Increased expression of miR-7 was found to suppress GBM cell proliferation, induce apoptosis, inhibit cell migration in vitro, and reduce tumorigenicity in vivo.4 However, in contrast to our study, other research group showed miR-7 downregulation in patients with GBM.18 Possible explanation for different outcomes might be in different normalization strategies used for qPCR (i.e. different reference genes used). miR-9 over-expression, observed in our primary and recurrent GBM samples, might contribute to tumor progression since it has been identified that up-regulation of miR-9 predicts an unfavorable prognosis and correlates with tumor progression.19 Similarly to other cancer research20, we observed increased expression of oncogenic miR-21 in primary and recurrent GBM, although its’ up-regulation in recurrent tumors is significantly lower compared to primary tumors. miR-26b was also found to be mainly over-expressed, although there are samples with its downregulation, but in contrast to our results, it has been shown that miR-26b was decreased in glioma cell lines and inversely correlated with the grade of glioma in humans.21 The discrepancy could be due to different normalization strategies. We normalized expression of miRNAs in GBM samples to Human Brain Reference RNA, in others the normal brain tissue was used for normalization.21 Expression of miR-124a in our research showed up-regulation in certain samples of primary tumors. Published data showed decreased miR-124a expression in majority of GBM samples due to hyper-methylation of promoter region of miR-124a.22 Speculatively, different methylation status of miR-124a might be observed in our samples. We also observed mainly higher expression level of miR-199a in primary GBM samples and near to normal level in recurrent GBM. miR-199a up-regulation in gliomas has been related to hypo-methylation of one of two copies of miR-199a within genome consequently leading to its up-regulation.14miR-199a is believed to have diverse biological functions in different tissues. It behaves as a tumor-suppressor in certain cancer types14; however, its exact role in GBM is yet to be defined. miRNA of let-7 family (let-7a/b/d/f) showed the most heterogeneous expression across both, primary and recurrent tumor samples.15,23

Our data also showed statistically significant alteration of expression of seven miRNAs (miR-7, miR-9, miR-21, miR-26b, miR-124a, miR-199a and let-7f) between primary and recurrent tumors. After receiving RT and ChT, we observed higher levels of miR-7 in recurrent compared to the primary GBM, whereas in patients treated solely using RT there was no significant difference in expression. Based on glioma cell line models, up-regulation of miR-7 increases radio-sensitivity.24 Several target genes that might be involved in sensitivity of glioma cells to therapy has been shown for miR-7, including EGFR, IRS1 and IRS2 regulating differentiation, invasion and proliferation4, as well as IGF-1R/Akt signalling pathway, regulating cellular growth and glucose metabolism in gliomas.25,26 We did observe a decreased expression of miR-9 in recurrent compared to primary GBM, however, miR-9 is still over-expressed in recurrent GBM. miR-9 has a role in chemotherapeutic resistance by regulating the p38 mitogen activated protein kinase signaling network, involved in regulation of cell stress and cell differentiation; it influences proliferation and self-renewal of glioma cells by targeting SOX2; and through miRNA regulation of PTCH1, receptor in Sonic Hedgehog signaling.26miR-21 has been shown to be up-regulated in more radio-resistant cell lines 12 mediating the resistance of GBM cells against RT27 and TMZ.20miR-21 is substantially implicated in pathogenesis of GBM by targeting PTEN, RECK, PDCDC4, TIMP3, BCL2, SPRY2, MTAP, SOX5, JMY, TGFRBR2, TGFBR3, TP73L, APAFI, BMPR2, TOPORS, DAXX, TP53BP2, and PPIF. All this target genes are involved either in pathways of apoptosis, invasion, proliferation and tumor growth either in chemo-resistance.4 Therefore, all these results suggest that lower expression of miR-21 after treatment might implicate less resistant glioma cells to therapy; however, as well as miR-9, it is still up-regulated in recurrent GBM, possibly contributing to tumor progression.

Decrease expression of miR-26b in our recurrent samples compared to primary GBM, could be a consequence of tumor progression. Accordingly, it was found that ectopic expression of miR-26b inhibits proliferation, migration and invasion of human glioma cells by repressing the endogenous level of EphA2 protein.21 We further observed decrease in miR-124a expression in recurrent tumors compared to primary, which might be a consequence of tumor progression. It has been shown that miR-124a radio-sensitizes human glioma cells by targeting CDK4.28 Several other targets have been also demonstrated for miR-124a, as are CDK6, SCP1, PTBP1, ITGB1, LAMCI, IQGAPI, RB and STAT3, regulating cell cycle, differentiation, invasion and proliferation of glioma cells and T cell-mediated immune clearance of glioma.4,29 All these processes may be involved in sensitivity and survival of glioma cells.

In recurrent GBM, but not in primary GBM, we found significant change in expression between samples of patients who were treated with RT and ChT and those that were treated with RT alone for only miR-15b, let-7d and let-7f. Let-7f was over-expressed in recurrent compared to the primary GBM; however, it was up-regulated only in patients treated with RT after first surgery. It has been shown that members of let-7 miRNA family (including let-7f) were mostly up-regulated in more radio-resistant cell lines12 suggesting radio-resistant function in glioma cells, since its expression after receiving RT and ChT was back to normal. There was observed normal expression of let-7d in GBM patients after receiving RT, however, after treating patients with RT and ChT, its expression was decreased and similar to that in primary GBM. There are limited data about the role of let-7d in the cell response to radiation and chemotherapeutic drugs in glioma, since depends on dose, time after irradiation and source of oxidative stress.30 There are indication that let-7d down-regulation in GBM might be involved in modulation of chemo-resistance to TMZ.11 One of the target genes of let-7d is oncogene HMGA231 and studies suggest that overexpression of HMGA2 may participate in regulating tumor cell invasion in high grade gliomas.32 However, whether HMGA2 is up-regulated in GBM due to let-7d down-regulation, is yet to be defined. Another miRNA, miR-15b, was found in our study to be significantly under-expressed in patients treated with RT and ChT compared to those treated with RT alone. Similarly, to let-7d, decrease of miR-15b did not change in recurrent GBM compared to primary GBM. miR-15b was demonstrated to suppress proliferation and induce apoptosis in glioma cells by targeting CCNE1.33miR-15b expression decreases with the ascending histopathological grade and its low expression is associated with poor overall survival of patients with high grade gliomas.34 Already known functions about radio-and chemoresistance/sensitivity and analyzed miRNAs are summarized in Figure 6.

FIGURE 6

Finally, in our patients we have observed longer disease free survival in patients treated with RT and ChT after first surgery compared to those treated with RT alone. The standard care treatment of GBM altered importantly after year 2005 when RT alone was replaced by RT associated with chemotherapeutic TMZ in great majority of our patients. Similarly to our results, it has been shown that for patients treated with surgery and RT alone median survival was shorter (12–12.1 months) compared to patients treated additionally with TMZ (14.2–14.6 months).35

Investigating miRNAs expression and survival in patients with GBM revealed that, both, miR-26b and let-7f represent better prognosis when downregulated. Lower expression of miR-26b is in correlation to progression free survival, whereas lower expression of let-7f is in correlation to overall survival. To the best of our knowledge, this is the first report describing expression of these two miRNAs to the function of survival in the GBM.

Although we did not found any differential expression of investigated miRNAs between methylated and unmethylated MGMT, we did observed change in MGMT methylation status from primary to recurrent GBM. Eight of nine unmethylated samples in primary GBM were methylated in recurrent GBM. This change in methylation status of MGMT is not surprising, since there are several publications describing similar percentage of change; however all changes were from more to less favorable state for tumor progress ion.36, 37, 38, 39