SEARCH

SEARCH BY CITATION

Keywords:

  • glioblastoma;
  • cancer stem cells;
  • CD133;
  • tumor recurrence

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

BACKGROUND:

Experimental data suggest that glioblastoma cells expressing the stem cell marker CD133 play a major role in radiochemoresistance and tumor aggressiveness. To date, however, there is no clinical evidence that the fraction of CD133-positive cells in glioblastoma that recurs after radiochemotherapy may be relevant for prognosis.

METHODS:

The authors used immunohistochemistry to assess CD133 expression in 37 paired glioblastoma samples, including 1 primary tumor sample and 1 recurrent tumor sample, after patients received adjuvant radiochemotherapy. To assess the actual composition of the CD133-positive glioblastoma cell population, fluorescence-associated cell sorting (FACS) analysis was used to sort CD133-positive/CD45-negative cells that were assayed for tumor-specific chromosomal aberrations using interphase fluorescence in situ hybridization. To rule out endothelial precursor cells, CD133-positive fractions also were assayed with anti-CD34 by FACS.

RESULTS:

In recurrent glioblastomas, the percentage of CD133-positive cells was increased by 4.6-fold compared with the percentage in primary glioblastomas, although, in some tumors, it increased up to 10-fold and 20-fold. Unexpectedly, the increase in CD133 expression was associated significantly with longer survival after tumor recurrence. An analysis of tumor-specific chromosomal aberrations and in vivo studies revealed that the CD133-positive cell compartment of recurrent glioblastoma was composed of both cancer stem cells and nontumor neural stem cells. The latter cells represented from 20% to 60% of the CD133-positive cell population, and their relative percentage favorably affected the survival of patients with recurrent glioblastoma. Endothelial CD133-positive/CD34-positive precursors did not contribute to the CD133-positive cell population.

CONCLUSIONS:

The authors hypothesized that, similar to the phenomenon described in glioblastoma models, neural stem/progenitor cells that are recruited by the tumor from surrounding brain may exert an antitumorigenic effect. Cancer 2011. © 2010 American Cancer Society.

Despite aggressive multimodal therapy, the prognosis for patients with glioblastoma remains very poor.1 Recently, tumor cells with stem-like features have been identified in glioblastoma.2, 3 These cells, which are referred to commonly as cancer stem cells (CSCs), express the transmembrane glycoprotein CD133 and have the ability to initiate tumor formation in vivo.2-4 It has been hypothesized that glioblastoma CSCs may arise from the malignant transformation of neural stem cells (SCs) that are resident in the subventricular zone and that the migration of these cells may lead to the development of glioma in different areas of the brain.5 Although tumorigenicity may be not confined entirely to the CD133-positive cell fraction of glioblastoma,6-9 growing data demonstrate that CD133 expression is related to an adverse prognosis10-14 and that the CD133-expressing glioblastoma cells contribute to radiochemoresistance and tumor aggressiveness.15-19 For example, Liu and colleagues demonstrated a positive selection for CD133-positive tumor cells in patients with glioblastoma after chemotherapy.15 The resistance of CD133-positive glioblastoma cells to cytotoxic drugs was related to high levels of drug transporter and inhibitor of apoptosis proteins and to enhanced methyl guanine-DNA methyl transferase (MGMT)-mediated DNA repair pathways.15 Bao and colleagues demonstrated that radiation treatment enriched glioblastoma xenografts with CD133-expressing cells and that radioresistant glioblastoma cells from irradiated xenografts formed secondary tumors with decreased latencies relative to untreated xenografts.19 In addition, CD133-positive cells that were isolated both from human glioma xenografts and from primary glioblastoma specimens activated the DNA damage checkpoint in response to radiation and repair radiation-induced DNA damage more effectively than CD133-negative tumor cells.19 Despite these results, there is no evidence that higher fractions of CD133-positive cells in glioblastoma that recurs after radiochemotherapy are correlated with a worse prognosis. It also remains to be determined whether and the extent to which the neural SCs that exhibit positive tropism both toward brain tumors and toward areas of brain inflammation20-22 may affect the composition of the CD133-positive cell fraction of glioblastoma.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Patient Selection

This study included 37 consecutive adult patients who underwent craniotomy for resection of histologically confirmed glioblastoma (World Health Organization grade IV)23 in the supratentorial compartment and who underwent reoperation for tumor recurrence after radiochemotherapy (Table 1). Both surgical procedures were performed at the Institute of Neurosurgery, Catholic University School of Medicine, Rome. All patients provided written informed consent according to the research proposals approved by the Ethical Committee of the Catholic University. Patients of pediatric age were not included. The patients were aged 28 to 74 years at the time of primary surgery (median age, 54 years; mean age [±standard deviation], 54.2 ± 10.3 years) and included 17 men and 20 women. In all patients, macroscopic total tumor resection was achieved both at primary surgery and at reoperation. After primary surgery, all patients received radiotherapy to limited fields (2 grays [Gy] per fraction, once daily 5 days a week; 60 Gy total dose) and adjuvant temozolomide.12 Overall survival was calculated from the date of primary surgery to the date of death; and survival after recurrence was calculated from the date of reoperation to the date of death.

Table 1. General Features of Patients and Tumor Biomarkers
Patient No.Age,yaSexKPSaKi-67, %aMGMTa,bOS, moSurvival After Disease Recurrence, moCD133, %CD133 Recurrence/ Primary Ratio
PrimaryRecurrence
  • KPS indicates Karnofsky performance status; MGMT, methyl guanine-DNA methyl transferase; OS, overall survival.

  • a

    Values refer to glioblastoma at primary surgery.

  • b

    See Pallini et al, 2008.12

167Woman7040Unmethylated9.02.0331.0
252Man7040Unmethylated9.03.5111.0
356Woman6020Unmethylated16.510.50.524.0
464Man8015Methylated16.06.010101.0
556Woman9025Unmethylated12.03.0111.0
654Woman805Unmethylated16.04.0310.3
751Man9025Unmethylated14.05.0155.0
849Woman9015Methylated34.07.00.524.0
950Man7045Methylated11.02.0155.0
1044Woman8010Methylated24.015.0133.0
1156Man9015Unmethylated33.011.0155.0
1243Man9020Methylated28.07.00.536.0
1336Woman7040Unmethylated8.04.010.50.5
1462Man8020Methylated14.02.0881.0
1565Woman8020Unmethylated19.09.0551.0
1668Woman8050Unmethylated9.03.031.50.5
1766Man8040Methylated10.03.00.50.51.0
1842Woman9015Unmethylated21.06.0133.0
1953Man7025Unmethylated24.08.0155.0
2057Man9030Unmethylated18.09.011010.0
2128Woman9025Unmethylated13.07.0111.0
2250Man8030Methylated16.05.011010.0
2353Man8020Unmethylated27.018.00.51020.0
2456Man9030Methylated20.011.011515.0
2552Woman8040Methylated32.08.0252.5
2654Man7070Unmethylated11.03.0221.0
2762Man9020Methylated28.013.01.542.6
2848Man9025Methylated51.020.0133.0
2974Woman9030Methylated29.013.04102.5
3041Woman7040Unmethylated9.52.0111.0
3144Woman7035Unmethylated12.03.0111.0
3266Woman7030Methylated11.04.0750.7
3342Woman9020Methylated42.09.0177.0
3448Woman8045Unmethylated9.03.0320.7
3567Woman7035Methylated10.04.0221.0
3658Man6040Unmethylated8.03.0310.3
3770Woman8025Methylated19.010.010151.5

Immunohistochemistry and Immunofluorescence

Immunohistochemistry was performed on deparaffinized sections using the avidin-biotin-peroxidase complex method (ABC-Elite Kit; Vector Laboratories, Burlingame, Calif).24 Endogenous biotin was saturated with a biotin-blocking kit (Vector Laboratories). For antigen retrieval, paraffin sections were microwave-treated in 0.01 M citric acid buffer, pH 6.0 (2 cycles for 5 minutes each at 750 W), followed by inhibition of endogenous peroxidase with 3% H2O2 for 5 minutes. Then, the sections were incubated with antibody directed against CD133 (dilution 1:50; Biocare Medical, Concord, Calif). After a 1-hour incubation at room temperature, immunodetection was performed using goat antirabbit secondary antibody (Vector Laboratories) and freshly made diaminobenzidine as a chromogen. For human telomerase reverse transcriptase (hTERT) antigen retrieval, paraffin sections were microwave-treated in ethylene diamine tetracetic acid buffer, pH 8.0, for 10 minutes. The percentage of CD133-positive cells was evaluated independently by 2 pathologists (F.P. and L.M.L.) who were unaware of the clinical data. For each slide, a minimum of 10 nonsuperimposing fields was examined at high-power magnification (×400), and at least 2000 tumor cells were counted in areas that were devoid of necrosis, hemorrhage, and abundant new vessel formation. Interobserver agreement was reached in the first analysis in 90% of slides; for the remaining slides, a consensus was reached by a joint review of the slides. For immunophenotyping of sorted cells, the cells were fixed with 4% paraformaldehyde and stained with antibody directed against nestin (Chemicon, Temecula, Calif), sex-determining Y-box 2 (SOX2) (R&D Systems, Minneapolis, Minn), musashi homolog 1 (Musashi1) (R&D Systems), glial fibrillary acidic protein (GFAP) (Dako, Glostrup, Denmark), and hTERT (Novocastra Laboratories, Newcastle upon Tyne, United Kingdom). Appropriate secondary antibodies (affinity-purified goat antimouse tetra methyl Rhodamin isothiocyanate-conjugated immunoglobulin G [IgG] and goat antirabbit fluorescein isothiocyanate [FITC]-conjugated IgG; both from Chemicon) were used.

Flow Cytometry and Fluorescence-Activated Cell Sorting for CD133, CD45, and CD34

Glioblastoma cells obtained from freshly dissociated surgical specimens were washed and resuspended in phosphate-buffered saline (PBS) plus 0.5% bovine serum albumin (BSA) (PBS-BSA) then incubated with anti-CD133 phycoerythrin (PE)-conjugated antibody (clone AC133-PE, mouse IgG1; Miltenyi Biotec, Bergisch Gladbach, Germany) and with either anti-CD45 FITC-conjugated antibody (clone T29/33-FITC, mouse IgG1; DakoCytomation, Carpenteria, Calif) or anti-CD34 FITC-conjugated antibody (clone BIRMA-K3 mouse IgG1; DakoCytomation) antibody. PE-conjugated and an FITC-conjugated mouse IgG1 isotype antibodies were used as controls according to the manufacturer's instructions for 60 minutes at 4°C. Then, the cells were washed with 2 mL of PBS-BSA, centrifuged at 800 revolutions per minute (rpm) for 5 minutes, resuspended in 0.5 mL PBS-BSA, and analyzed by flow cytometry. Viable cells were identified using 7-amino actinomycin D. Cell sorting (BD fluorescence-associated cell sorting [FACS] Aria; Becton-Dickinson, Franklin Lakes, NJ) was performed using the same staining method described above for CD133-PE. Sorted cells were analyzed for purity by flow cytometry with a FACSCanto flow cytometer (Becton-Dickinson) and FACS Diva software (Becton-Dickinson). Sorted cells were fixed in a solution of methanol and acetic acid (3:1) for 10 minutes and then processed for fluorescence in situ hybridization (FISH) as described below.

Fluorescence In Situ Hybridization

Single-probe and dual-probe FISH was performed either on cell nuclei extracted from paraffin-embedded sections of the tumor or on the CD133-positive/CD45-negative and CD133-positive/CD45-positive sorted cells using locus-specific probes for the centromere of chromosome enumeration probe 10 (CEP 10) for telomere of chromosome 19 (tel 19q) and for a locus-specific probe on chromosome 22 (LSI22) (breakpoint cluster region locus q11.2; Vysis Inc. Abbot Laboratories SA, Downers Grove, Ill), as described elsewhere.25 These probes were chosen because loss of euploidy for chromosomes 10, 19, and 22 has been observed frequently in glioblastoma.12, 26, 27 Briefly, for nuclei extraction, 40-μm-thick sections were dewaxed with xylene, manually disaggregated, and digested with freshly prepared 0.005% proteinase K in TRIS 0.05 M, pH 7, for 30 minutes at 37°C. To aid with enzymatic digestion, the samples were vortexed for 5 seconds at 5-minute intervals during this incubation. Nuclei were pelleted using a microcentrifuge (6000 rpm) for 10 minutes. Proteinase K was carefully removed, and the nuclei were washed by resuspension with vortexing in 100 μL PBS. The PBS solution was removed, and the nuclear suspension was fixed in a solution of methanol and acetic acid (3:1). Eight microliters of nuclear suspension were placed on a slide that was positively charged within a 13-mm circle. The slides were then dried in a 65 °C oven for 15 minutes and treated in a microwave oven for 10 minutes in citrate buffer, pH 6.0, followed by enzymatic digestion with 4 mg/mL of pepsin in NaCl 0.9%, pH 1.5, for 20 minutes at 37 °C.28 Samples were then dehydrated in a graded ethanol series and subjected to FISH analysis. Sorted cells had been fixed previously in a solution of methanol and acetic acid (3:1) for 10 minutes. After specimen/probe denaturation at 73°C for 5 minutes, the probe (10 μL to the slide) was applied to the slides, and the slides were incubated at 42°C overnight for CEP10 and at 37 °C for 10 to 16 hours for LSI22/tel 19q. The posthybridization procedure included subsequent washing in 50% formamide/2 × standard saline citrate (SSC) (for 30 minutes at 46°C) and in 2 × SSC 0.1% NP40 (for 5 minutes at room temperature). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI) (Vectashield mounting medium with DAPI; catalog number H-1200; Vector Laboratories). The slides were evaluated with an Axioplan fluorescent microscope (Karl Zeiss, Gottingen, Germany) equipped with appropriate filters (Vysis Inc.). In each hybridization, at least 500 cells in interphase were analyzed.

Intracranial Grafting of Glioblastoma Cancer Stem Cells in Immunodeficient Mice

Severe combined immunodeficient (SCID) mice (both sexes, aged 4 weeks; Charles River, Lecco, Italy) were used. Mouse research was approved by the Ethical Committee of the Catholic University School of Medicine in Rome. For intracranial grafting, 3 × 104 FACS-sorted CD133-positive/CD45-negative glioblastoma cells were resuspended in 4 μL Dulbecco modified Eagle medium. The mice were anesthetized (diazepam 2 mg/100 g intraperitoneally followed by ketamine 4 mg/100 g intramuscularly), the skull was immobilized in a stereotactic head frame, and a burr hole was made 2 mm to the right of the midline and 1 mm behind the coronal suture. The tip of a 10-μL Hamilton microsyringe was placed at a depth of 3.5 mm from the dura, and the cells were slowly injected. After grafting, mice were monitored and killed 16 weeks after grafting or when they presented with signs of tumor. The brain was removed, fixed in 4% paraformaldehyde, embedded in paraffin, and cut on the coronal plane either in 5-μm-thick sections for histology or in 40-μm-thick sections for cell nuclei extraction. The 5-μm-thick sections were stained with hematoxylin and eosin. Nuclei extraction was performed as described above.

Statistical Analysis

The percentages of CD133-positive cells in glioblastoma at first diagnosis and at recurrence after radiochemotherapy were compared using the Student t test. Correlations between the expression of CD133 and either overall survival or survival after recurrence were studied using regression analysis and the Pearson correlation coefficient. Curves for overall survival and for survival after tumor recurrence were estimated by using the Kaplan-Meier method and were compared by using 2-sided log-rank tests. A Cox proportional-hazards model was fitted to assess the prognostic value of CD133 expression, Ki67 expression, MGMT methylation, and other potential prognostic factors, like patient age and preoperative performance status,29, 30 by generating different models that considered overall survival. Correlation analyses between variables were performed before including each variable in each model using MedCalc software, version 9.5.0 (MedCalc Software, Mariakerke, Belgium). The results are reported as 2-sided P values with 95% confidence intervals (CIs).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Increased CD133 Expression in Recurrent Glioblastoma and Its Relation to Survival

By using immunohistochemical staining, we assessed the percentage of CD133-positive cells in 37 paired glioblastoma tumors both at first diagnosis before any therapy and at tumor recurrence after surgery and adjuvant radiochemotherapy (Fig. 1, Table 1). In each patient, any change in CD133 expression was calculated as the CD133 ratio (ie, the ratio between the percentage of CD133-positive cells in recurrent glioblastoma and the percentage of CD133-positive cells in glioblastoma at first diagnosis). Compared with glioblastoma at first diagnosis, in recurrent tumors, the CD133-positive cell population was increased in 19 patients (51.3%; CD133 ratio, >1), unchanged in 12 patients (32.4%; CD133 ratio, 1), and reduced in 6 patients (16.2%; CD133 ratio, <1) (Fig. 1C). The mean CD133 ratio was 4.6; however, in some instances, it reached values of 10 to 20 (Table 1). Overall, the mean percentage (±standard error of the mean [SEM]) of CD133-positive cells was 2.34% ± 0.42% (range, 0.5%-10%) and 4.55 ± 0.65 (range, 0.5%-15%) in glioblastoma at first diagnosis and at recurrence after radiochemotherapy, respectively (P < .001; Student t test) (Fig. 1D).

thumbnail image

Figure 1. (A) These are gadolinium-enhanced, axial, T1-weighted magnetic resonance images of a glioblastoma from Patient 24 (Left) before surgery, (Middle) after surgery and adjuvant radiochemotherapy (Rxtherapy), and (Right) at the time of tumor recurrence before reoperation. (B) Anti-CD133 immunohistochemistry in the same tumor (Left) at primary surgery and (Right) at recurrence after radiochemotherapy reveals an increase in the number of immunostained cells in the recurrent tumor. Scale bar = 50 μm. Insets: Details of CD133-positive cells are shown at higher magnification. (C) Expression of the transmembrane glycoprotein CD133, as assessed by the ratio between the percentage of CD133-positive cells in glioblastomas that recurred after radiochemotherapy and the percentage of the same cells in primary tumors, was increased in 19 tumors (CD133 ratio, >1), was unchanged in 12 tumors (CD133 ratio, 1), and was decreased in 6 tumors (CD133 ratio, <1). (D) The percentage of CD133-positive cells, as averaged from all patients, increased significantly in recurrent glioblastomas compared with primary glioblastomas (P < .001; Student t test).

Download figure to PowerPoint

The increased CD133 expression in glioblastoma that recurred after radiochemotherapy was not an unfavorable factor for prognosis; in fact, it was associated significantly with longer survival after tumor recurrence (Fig. 2). It is interesting to compare linear regressions that analyzed overall survival and percentages of CD133-positive cells in primary tumors and recurrent tumors (Fig. 2A,B). Although the phenomenon did not reach statistical significance, in primary glioblastomas, the percentage of CD133-positive cells was associated with shorter survival (Fig. 2A). In recurrent glioblastomas, however, the CD133-positive cell fraction appeared to acquire an opposite association (Fig. 2B,C). When survival after recurrence was considered in a regression analysis, then both the percentage of CD133-positive cells in recurrent tumor and the CD133 ratio were associated significantly with a better prognosis (Fig. 2C,E). The favorable prognostic value of increased CD133 expression in recurrent glioblastoma is demonstrated clearly by the survival curves for patients who had tumors with a CD133 ratio >1 (n = 18) relative to patients who had tumors with a CD133 ratio ≤1 (n = 19; P ≤ .0001 for both overall survival and survival after recurrence; log-rank test) (Fig. 3A,B). On multivariate analysis, a CD133 ratio >1 (P = .00013; 95% CI, 0.0394-0.3489), Ki-67 expression (P = .03365; 95% CI, 1.0024-1.0571), and Karnofsky performance status (P = .04635; 95% CI, 0.8981-0.9989) emerged as independent predictors of overall survival. Age was not related significantly to patient survival; however, this result does not necessarily suggest that our series is not a representative study sample, because it may depend on the exclusion of older patients for reoperation.

thumbnail image

Figure 2. These graphs illustrate linear regression analyses of CD133 expression and survival. (A-C) Graphs plot overall survival and survival after tumor recurrence in relation to (A) the percentage of CD133-positive cells at primary surgery before (Pre) radiochemotherapy (Rx-ChemoTherapy), (B,C) the percentage of CD133-positive cells at recurrence after (Post) radiochemotherapy, and (D,E) the ratio of CD133-positive cells at recurrence to CD133-positive cells at primary surgery. The relation between variables was assessed with 95% confidence intervals using regression analyses and Pearson linear correlation coefficients (P).

Download figure to PowerPoint

thumbnail image

Figure 3. These are probability curves for (A) overall survival and (B) survival after tumor recurrence in 37 patients who had glioblastomas with different levels of CD133 expression and who underwent surgery and received adjuvant radiochemotherapy. There was a highly significant difference in both overall survival and survival after recurrence between the curves for patients who had glioblastomas in which CD133 levels increased after radiochemotherapy (CD133 ratio, >1) and patients who had glioblastomas in which CD133 levels were unchanged or reduced (CD133 ratio, ≤1).

Download figure to PowerPoint

Heterogeneous Nature of the CD133-Positive Cell Population in Recurrent Glioblastoma

To explain why the increased CD133 expression in recurrent glioblastoma (after radiochemotherapy) was associated with longer survival, we hypothesized that, in recurrent glioblastoma, the CD133-positive cell population may not be constituted entirely by tumor cells and that it also may contain nontumor, stem-like cells. Experiments using homotopic human glioma xenografts have demonstrated a remarkable tropism of endogenous neural SCs toward the area where the tumor had been implanted.20, 21 In addition, postoperative glial scarring and radiotherapy may enhance the migration of endogenous neural SCs cells in the recurrent tumor.22 Therefore, in a subset of recurrent glioblastomas (n = 17), we used FACS analysis to sort the CD133-positive cells from freshly dissociated specimens. To rule out CD133-positive cells of hematopoietic origin,31 the cells also were labeled with anti-CD45 (Fig. 4A,C). Sorted CD133-positive/CD45-negative and CD133-positive/CD45-positive cells were then assessed with an interphase FISH assay using probes for chromosomes 10, 19, and 2225 (Tables 2 and 3). We observed that each glioblastoma that harbored aneuploidy for 1 or more of these chromosomes had a fraction of euploid CD133-positive/CD45-negative cells that which ranged from 20% to 60% (mean ± SEM, 38.1% ± 5.7%) (Fig. 4B,C; Table 3). In any given tumor, all CD133-positive/CD45-positive cells were euploid (Fig. 4B). To rule out the possibility that endothelial precursor cells were involved in tumor angiogenesis,32, 33 CD133-positive fractions also were assayed for CD34 by FACS analysis. In each tumor, however, the CD133-positive/CD34-positive cells were so rare that they could not substantially affect the composition of the CD133-positive cell fraction (Fig. 5).

thumbnail image

Figure 4. Interphase fluorescence in situ hybridization (FISH) assays were used to analyze tumor-specific chromosomal changes in CD133-positive (CD133+) cells from patients with recurrent glioblastoma. (A) (Left) The histologic picture of recurrent glioblastoma in Patient 7 is shown (Harris hematoxylin [H&H] stain; scale bar = 200 μm). Fluorescence-activated cell sorting (FACS) analyses were done on freshly dissociated tumor for (Middle) immunoglobulin G-phycoerythrine (IgG-PE) and IgG-fluorescein isothiocyanate (IgG-FITC) isotype control antibodies (IgG-FITC-A) and for (Right) CD133/CD45 expression. In this recurrent glioblastoma, the proportion of CD133-positive/CD45-negative (CD45−) cells was 5.1%. (B) A FISH assay was used to assess cell nuclei that were extracted from paraffin sections of (Left) glioblastoma cells, (Middle) sorted CD133-positive/CD45-negative cells, and (Right) sorted CD133-positive/CD45-positive cells. In this tumor, 70% of CD133-positive/CD45-negative cells revealed monosomy for chromosome enumeration probe 10 (CEP10), as did the tumor cells, whereas the remaining 30% of CD133-positive/CD45-negative cells were euploid for CEP 10, and CD133-positive/CD45-positive cells also were euploid. (C) (Left) The histologic picture of recurrent glioblastoma in Patient 16 is shown (Harris hematoxylin stain; scale bar = 200 μm). (Middle) FACS analysis revealed 1.4% CD133-positive/CD45-negative cells (GPA indicates glycophorin A). (Right) A FISH assay that analyzed sorted CD133-positive/CD45-negative cells revealed either tumor-specific amplification of telomere-like primer 19 (Tel19) or normal diploid FISH signal.

Download figure to PowerPoint

thumbnail image

Figure 5. Fluorescence-activated cell sorting (FACS) analysis was used to analyze a freshly dissociated specimen of recurrent glioblastoma for (Left) immunoglobulin G-phycoerythrine (IgG-PE) and IgG-fluorescein isothiocyanate (IgG-FITC) isotype control antibodies (A) and for (Right) CD133/CD34 expression. The fraction of CD133-positive/CD34-positive cells is so small that it cannot distinguish reliably from FACS analysis with control antibodies.

Download figure to PowerPoint

Table 2. Results of Fluorescence In Situ Hybridization Assays on Cell Nuclei Extracted From Paraffin Sections of Recurrent Glioblastoma
Patient No.FISH
Cep10Tel19LSI22
  1. FISH indicates fluorescence in situ hybridization; Cep10, chromosome enumeration probe 10; Tel19, telomere-like primer 19; LSI22, locus-specific probe on chromosome 22.

1NormalNormalNormal
2MonosomyNormalMonosomy
5NormalNormalNormal
6NormalNormalNormal
7MonosomyNormalNormal
9NormalNormalNormal
11NormalNormalNormal
16PolysomyPolysomyPolysomy
24NormalNormalNormal
25NormalPolysomyNormal
26NormalNormalNormal
27PolysomyNormalNormal
28NormalNormalNormal
29MonosomyNormalNormal
33MonosomyNormalNormal
35MonosomyNormalNormal
36NormalNormalNormal
Table 3. Fluorescence In Situ Hybridization Analysis on Fluorescence-Activated Cell Sorting in Cells From Recurrent Glioblastoma
Patient No.FACSFISH
CD133+/ CD45− Cells, %Chromosomal AberrationEuploid CD133+/ CD45− Cells, %
  1. FACS indicates fluorescence-activated cell sorting; FISH, fluorescence in situ hybridization; +, positive; −, negative; Cep10, chromosome enumeration probe 10; LSI22, locus-specific probe on chromosome 22; Tel19, telomere-like primer 19.

21.2Cep10 monosomy, LSI22 monosomy20
75.1Cep10 monosomy30
161.4Cep10 polysomy, Tel19 polysomy, LSI22 polysomy30
254.5Tel19 polysomy60
273.6Cep10 polysomy45
2912.4Cep10 monosomy60
337.3Cep10 monosomy40
353.1Cep10 monosomy20

Because, in our series of recurrent glioblastomas, the percentage of CD133-positive cells that lacked chromosomal changes (38.1%) was higher than that reported by Bao and colleagues in 1 glioblastoma at first diagnosis (approximately 15%),19 we investigated whether this discrepancy had a technical explanation or reflected a specific feature of the recurrent tumors. Therefore, we used FACS analysis to sort out CD133-positive/CD45-negative cells from 3 primary glioblastomas (from Patients 27, 29, and 35); then, we assessed these cells for tumor-specific chromosomal changes using an interphase FISH assay. In primary glioblastomas, the percentage of CD133-positive/CD45-negative cells that did not harbor tumor-specific chromosomal aberrations ranged from 5% to 15% (mean, 9%). This result suggests that the euploid CD133-positive/CD45-negative cells increased remarkably in recurrent tumors.

We also assessed the phenotype of sorted CD133-positive/CD45-negative cells with immunocytochemistry, and we observed that the majority of these cells expressed the SC marker nestin (mean ± SEM, 61.2% ± 4%; range, 49%-78%) (Fig. 6A,B). A proportion of the CD133-positive/CD45-negative cells were not immunostained with either antinestin or anti-GFAP antibodies (mean ± SEM, 22.5% ± 2.6%; range, 13%-33%), whereas a minor fraction of these cells coexpressed both nestin and GFAP (mean ± SEM, 15.1% ± 2.7%; range, 6%-29%). The stem cell markers SOX2 and Musashi1 were expressed by 73.2% and 62.5% of the sorted CD133-positive/CD45-negative cells, respectively (Fig. 6C). Expression of hTERT, a feature that distinguishes tumorigenic human adult neural SCs cells from their normal counterparts,34 was absent in approximately 33% of CD133-positive/CD45-negative cells (mean ± SEM; 32.2% ± 6.4%; range 18%-47%) that sorted from hTERT-positive recurrent glioblastoma (Fig. 6D). This result provides additional evidence that a fraction of CD133-positive/CD45-negative cells in recurrent glioblastoma lacks the neoplastic features of the parent tumor.

thumbnail image

Figure 6. Immunophenotypes of sorted cells are illustrated. (A) CD133-positive/CD45-negative (CD133+/CD45−) sorted cells were analyzed further by antinestin and antiglial fibrillary acidic protein (anti-GFAP) immunocytochemistry. The majority of the sorted cells are stained only for nestin (Nest), suggesting an undifferentiated stem cell phenotype. The fraction of CD133-positive/CD45-negative cells that coexpress nestin and GFAP (arrows) may represent either CD133-positive neural precursors or cancer stem cells at different stages of differentiation. DAPI indicates 4,6-diamidino-2-phenylindole. (B) These are results from a comparison of fluorescence in situ hybridization (FISH) and immunocytochemistry analyses in sorted CD133-positive/CD45-negative cells. (C) Immunofluorescence of sorted CD133-positive/CD45-negative cells is shown with the stem cell markers sex-determining Y-box 2 (SOX2) and musashi homolog 1 (Musashi1). (D) Shown are (Left) the histologic picture and (Middle) and antihuman telomerase reverse transcriptase (anti-hTERT) immunohistochemistry in a glioblastoma from Patient 16. Scale bar = 50 μm. (Right) Anti-hTERT immunostaining for sorted CD133-positive/CD45-negative cells shows 3 cells without nuclear expression of hTERT. H&H indicates Harris hematoxylin stain.

Download figure to PowerPoint

Tumor and Nontumor CD133-Expressing Cells in Recurrent Glioblastoma

Glioblastomas are genetically heterogeneous neoplasms that are likely to acquire several mutations during the course of tumor progression. Therefore, the CD133-positive/CD45-negative cells sorted from recurrent glioblastomas that lacked aneuploidy for chromosome 10, 19, or 22 may have been tumor cells harboring mutations that were not detectable with the FISH probes used in our study. Alternatively, such euploid CD133-positive/CD45-negative cells may have been nontumor cells with stem-like features. To address this issue, in immunocompromised mice, we grafted CD133-positive/CD45-negative cells that were sorted from 3 recurrent glioblastomas in which FISH analysis had demonstrated aneuploidy for chromosome 10, 19, or 22 (Table 2, Fig. 7). In this experiment, the grafted cells were a mixed population that contained both euploid and aneuploid CD133-positive/CD45-negative cells. It is known that the ability to generate tumors in vivo is a landmark for glioblastoma CSCs that distinguishes these cells from normal neural SCs. If all of the CD133-positive/CD45-negative cells were neoplastic cells, the we expected that the tumor xenografts would contain both aneuploid and euploid cells. Five of 9 mice that were grafted with CD133-positive/CD45-negative cells developed highly infiltrating brain tumors that involved the entire striatum and piriform cortex with swelling of the hemisphere and compression of the lateral ventricle (Fig. 7). When the cell nuclei of the tumor xenografts were assessed by FISH using probes for human chromosome 10, 19, or 22, we observed that virtually all cell nuclei with positive FISH signals were aneuploid (Fig. 7), indicating that the euploid fraction of CD133-positive/CD45-negative cells sorted from recurrent glioblastoma was not tumorigenic. When we revisited the correlation between CD133 expression and survival in the 8 patients who had recurrent glioblastoma in which aneuploid CD133-positive/CD45-negative tumor cells could be dissected from euploid CD133-positive/CD45-negative stem-like cells, we observed that the relative presence of the latter cells was related significantly to a better outcome in terms of both overall survival and survival after recurrence (Fig. 8).

thumbnail image

Figure 7. The in vivo tumorigenicity of CD133-positive/CD45-negative cells is shown in recurrent glioblastoma. (A) Photomicrographs show (Left) the histologic picture (Harris hematoxylin stain; scale bar = 100 μm) and (Right) fluorescence in situ hybridization (FISH) analysis of fluorescence-activated cell sorted CD133-positive/CD45-negative glioblastoma cells from Patient 27. Trisomy for chromosome enumeration probe 10 (Cep10) is present in approximately 50% of CD133-positive/CD45-negative cells. Tel19 indicates telomerase-like primer 19. (B) (Top Left) A brain tumor xenograft developed 16 weeks after intracranial injection of CD133-positive/CD45-negative glioblastoma cells. (Top Right) The histologic picture of the tumor xenograft is shown (hematoxylin and eosin stain; scale bar = 25 μm). (Bottom Left) A FISH assay on cell nuclei that were extracted from paraffin sections of the tumor xenograft show Cep10 trisomy in all nuclei. (Bottom Right) This diagram compares the percentages of aneuploid cells in the CD133-positive/CD45-negative cell population sorted from the 3 parent glioblastomas and the tumor xenografts.

Download figure to PowerPoint

thumbnail image

Figure 8. These graphs plot the percentages of euploid CD133-positive/CD45-negative (CD133+/CD45−) cells both (Left) against overall survival and (Bottom) against survival after tumor recurrence. In the 8 patients who had recurrent glioblastoma, the relative percentage of euploid CD133-positive/CD45-negative cells in the recurrent tumor was related significantly to longer survival. P indicates Pearson linear correlation coefficient.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

In the current study, we assessed CD133 expression in a group of 37 glioblastomas that recurred after patients had undergone surgery and received adjuvant radiochemotherapy, and we observed that the percentage of CD133-positive cells was increased up to 20-fold compared with the percentage in the paired primary tumors. Surprisingly, the increased CD133 expression in recurrent glioblastoma was not an adverse factor for prognosis, because it often was associated with longer survival. Analyses of tumor-specific chromosomal aberrations and of hTERT expression in CD133-positive/CD45-negative cells that were sorted from recurrent glioblastomas as well as intracerebral xenografts of these cells in immunocompromised mice suggested that diverse subsets of CD133-positive cells contribute toward populating recurrent glioblastoma.

The heterogeneous composition of the CD133-positive cell population may help to explain some recent data that appear to diverge from the cancer stem hypothesis for glioblastoma. For example, Joo et al9 reported less dramatic clinical outcomes for patients who had glioblastomas with high CD133 expression (≥3%) compared with patients who had tumors that expressed low levels of CD133 (<3%), as determined by FACS analysis. Their result may be explained by the presence of nontumor CD133-positive cells in the surgical specimens. In a previous study by our group, we were unable to demonstrate a direct relation between the percentage of immunostained CD133-positive cells in glioblastoma and patient prognosis, whereas such relation emerged for proliferating CD133-positive/Ki67-positive cells.12 It is likely that the CD133-positive/Ki67-positive cells may represent the actual tumorigenic population of glioblastoma more closely than the CD133-positive cells as a whole, assuming that the proliferation of neural SCs is a very rare event.

Tumor and nontumor CD133-positive cells may play completely different roles in glioblastoma biology. In a mouse model, the glioblastoma-induced attraction of endogenous neural precursor cells was associated with improved survival because of antiproliferative and proapoptotic actions of the neural precursors in glioblastoma cells.35 Defining the actual role of the neural SCs in human glioblastoma and their interactions with CSCs was beyond the objectives of the current study. However, for those patients in whom the aneuploid CD133-positive CSCs could be distinguished from the euploid CD133-positive neural SCs, the relative presence of the latter cells was linked significantly to a better outcome.

In conclusion, CD133-positive cells in recurrent glioblastomas are a heterogeneous cell population composed of both CSCs that are likely to represent the actual tumorigenic fraction responsible for tumor progression and nontumor cells—putatively, neural SCs. The latter cells are supposed to migrate from surrounding brain toward the tumor in which they may exert an antiglioma effect. The CD133-positive endothelial precursor cells do not contribute substantially toward populating recurrent glioblastoma. These observations warrant further studies to distinguish the CD133-positive CSCs from nontumor, CD133-positive cells that are recruited by glioblastoma and to establish the actual role of the latter cells in tumor biology.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Supported by grants from Fondi d'Ateneo (D1) and Mrs. Paola Krajnik (to R.P.), and from Associazione Italiana per la Ricerca sul Cancro and the Italian Health Ministry (to R.D.M.).

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES