Article Text

The chemokine receptor CXCR7 influences prognosis in human glioma in an IDH1-dependent manner
  1. Peter Birner1,
  2. Andrey Tchorbanov2,
  3. Sevdalin Natchev3,
  4. Jochen Tuettenberg4,
  5. Marin Guentchev4
  1. 1Clinical Institute of Pathology, Medical University of Vienna, Vienna, Austria
  2. 2Department of Immunology, Institute of Microbiology, Bulgarian Academy of Sciences, Sofia, Bulgaria
  3. 3Department of Neuropathology, St. Ivan Rilski Hospital, Sofia, Bulgaria
  4. 4Department of Neurosurgery, Klinikum Idar-Oberstein, Idar-Oberstein, Germany
  1. Correspondence to Dr Marin Guentchev, Department of Neurosurgery, Klinikum Idar-Oberstein, Dr. Ottmar-Kohler Str. 2, Idar-Oberstein 55743, Germany; guentchev{at}


Aims The chemokine receptor CXCR7 is found on glioma cells and glioma-associated vessels and dependent upon its localisation on tumour or endothelial cells the CXCR7 receptor can mediate glioma cell invasion and tumour angiogenesis. Its expression predicts survival in several types of cancers.

Methods We immunohistochemically studied the expression of CXCR7 and its ligand SDF1α in a cohort of 354 human patients with glioma. In an in vivo glioma model, we studied the effect of selective CXCR7 inhibition on mean vascular density.

Results Here we show that expression of either mutant isocitrate dehydrogenase (IDH) 1 or podoplanin (PDPN), two proteins present in basically non-overlapping glioma populations, predicts the prognostic significance of CXCR7. Specifically, expression of CXCR7 on endothelial cells in IDH1 mutant cases predicted poor outcome. Surprisingly, in PDPN expressing gliomas, one of the marker genes for the recently identified mesenchymal subgroup, expression of CXCR7 predicts diminished prognosis on tumour cells and better prognosis on endothelial cells.

Conclusions Since CXCR7 is expressed on migrating cells our data suggest that, although ubiquitously present, angiogenesis and invasion are outcome-relevant events in specific glioma subgroups, providing a potentially important tool for targeted therapy assignment.


Statistics from


Large scale genetic profiling of human gliomas has identified at least three signatures of high-grade gliomas based on genetic and epigenetic patterns.1 A subgroup of gliomas with better prognosis is characterised by mutations in the genes encoding isocitrate dehydrogenase (IDH) 1 and 2.2 Another subgroup associated with the shortest survival was designated mesenchymal based on the dominant feature of its signature genes. One of the markers distinguishing the mesenchymal phenotype from all others is the cell surface protein podoplanin (PDPN).1

Despite intense research, it is still unclear how IDH1 and IDH2 influence prognosis in patients with cancer. Recent reports show that the mutated form inhibits the homodimerisation of the wild-type protein,3 blocks differentiation of non-transformed cells4 and is likely to be involved in tumorogenesis in part through induction of HIF-1 signalling.3 In a recent study, we were able to show that mutant IDH1 inhibits PI3K signalling, thus possibly delays tumour progression.5

The chemokine receptor CXCR7 and its ligand6 are expressed on both glioma cells (tcCXCR7, tcSDF1a) and tumour-associated vessels (vasCXCR7, vasSDF1a).7 Both molecules have been implicated in cancer angiogenesis8 ,9 and tumour cell migration,10 where CXCR7 was expressed on the migrating cell and SDF1a acted as a chemoattractant.11 ,12 A recent study has shown that in the rodent glioblastoma multiforme models, pharmacological inhibition of CXCR7 post irradiation caused tumour regression, blocked tumour recurrence and/or substantially prolonged survival.13

Here we show that expression of CXCR7 on endothelial cells in IDH1 mutant cases predicts poor outcome while in a non-overlapping glioma population expressing the cell surface protein PDPN, expression of CXCR7 on tumour cells predicts diminished prognosis.

Material and methods


Our discovery cohort represented 354 patients operated at St. I. Rilski Hospital used previously.5 ,14 Seventy-two cases were diagnosed with WHO grade II gliomas (mean age 40±14 years), 32 grade III cases (mean age 45±16 years) and 250 grade IV cases (mean age 56±13 years). Tissue microarrays were constructed containing one spot with 4 mm diameter per sample. Following antibodies were used for immunohistochemistry: PDPN (ready-to-use, clone D2-40, Cell Marque), IDH1R132H (clone H09, gift from Professor von Deimling), CXCR7 (clone 11G8, 1:50, R&D), SDF1α (clone 79018, 1:25, R&D), Ki67 (clone MM1, 1:20, Novocastra), CD31 (clone SZ31, 1:20, Dianova).


Overall survival was defined as the period from primary surgery until death, regardless of cause. Survival until the end of the observation period was considered a censored observation. Univariate analysis of overall survival was performed using log-rank analysis or univariate Cox regression in case of patients’ age. Any variables that had a p value ≤0.05 in the univariate analysis were carried forward into multivariable analysis. Multivariable analysis of survival was performed with the Cox proportional hazards model.

For Cox regression, adjusted HRs with 95% CI are presented. The Kruskal–Wallis test was used for comparisons between groups in the animal experiments. SPSS V.20.0 was used for all calculations. A two-tailed p value of ≤0.05 was considered as significant; due to the exploratory design of the study, no correction for multiple testing was performed.

Glioma angiogenesis model

For studying the role of CXCR7 in glioma angiogenesis, we chose a subcutaneous model for several reasons: (1) study focuses only on angiogenesis and not on the clinical effectiveness of the drug, (2) lack of information on how effectively the CXCR7 inhibitor used crosses the blood-brain-barrier and (3) lower mortality rates as compared with the intracranial model. In summary, 3×106 U373MG glioma cells (American Type Culture Collection, not genetically verified within the last 6 months) were injected into the flanks of 8-week-old to 12-week-old severe combined immunodeficiency mice and treated with the CXCR7 inhibitor CCX771 (ChemoCentryx) from day 3 to 24 post infection. All procedures were approved by the local institutional review board and the BAS Animal Care and Use Committee.


CXCR7 and SDF1a are found on both tumour cells and tumour endothelial cells

We evaluated the vascular (vas) and tumour cell (tc) presence of CXCR7 and SDF1α (figure 1). Expression of tcCXCR7 was found in 11.6% (37/320), vasCXCR7 in 58.1% (186/320), tcSDF1a in 35.7% (111/311) and vasSDF1a in 62.4% (194/311) of all cases (table 1). Both markers were found on tumour cells and tumour endothelial cells in all WHO grades. A hot spot Ki67 index of all cases was determined as well.

Table 1

Expression of proteins in relation to WHO grading and PDPN/mutIDH1 expression

Figure 1

Immunostaining for Ki67, CXCR7 and SDF1α. (A) Case positive for vasCXCR7, (B) case positive for tcSDF1α, (C) case positive for tcCXCR7 and (D) case with a high Ki67 index. Bar represents ∼150 μm.

Strong positive associations of tcSDF1a expression with WHO grade (p<0.001, χ2 test) and PDPN expression (p=0.002, χ2 test), and a negative association of tcSDF1a with IDH1 expression (p<0.001, χ2 test) were found. No such associations for vascSDF1a were found.

An association of vascCXCR7 with WHO grade (p<0.001, χ2 test) and PDPN expression (p<0.001, χ2 test), and a negative association of vascCXCR7 with IDH1 status (p<0.001, χ2 test) were found. We also found a significant association of tcCXCR7 with WHO grade (p=0.023, χ2 test) and PDPN expression (p=0.023, χ2 test), and a negative one with IDH1 status (p=0.048, χ2 test).

From these markers, only age, WHO grade, vascCXCR7 and PDPN had a statistically significant impact on survival in all cases (see online supplementary table S1).

Expression of mutant IDH1 and PDPN determines the prognostic significance of CXCR7 and SDF1a

In IDH1 mutant cases, expression of both tcSDF1α (p=0.001, HR 3.738; 95% CI 1.669 to 8.371) and vasCXCR7 (p=0.008, HR 2.969; 95% CI 1.336 to 6.598) predicted poor outcome. Interestingly, a high Ki67 index was also associated with shorter survival only in the mutIDH subgroup.

Since PDPN is predominantly found in the mesenchymal type of glioma1 and in our series almost never overlaps with mutIDH, PDPN+ cases were used as a control group. In these cases, the expression of tcCXCR7 predicted diminished prognosis (p=0.016, HR 1.912; 95% CI 1.128 to 3.24) and that of vasCXCR7 predicted a better (p=0.031, HR 0.582; 95% CI 0.356 to 0.952) prognosis (figure 2). All other markers did not show any prognostic significance in the PDPN+ glioma subgroup.

Figure 2

Kaplan–Meier curve representing the survival impact of CXCR7 and SDF1α. Expression of CXCR7 on endothelial cells (A) and SDF1α on tumour cells (B) in isocitrate dehydrogenase (IDH) 1 mutant cases predicted poor outcome. Interestingly, expression of CXCR7 on tumour cells (C) predicted diminished prognosis in PDPN+ gliomas. The numbers on the x-axis represent days. Note that vasCXCR7 has a contrary effect on survival in mutIDH1 and PDPN+ gliomas. The negative prognostic impact of vasCXCR7 in mutIDH1 cases is in concert with the effect of tumour cell presence of its ligand SDF1α.

Multivariable analysis of survival impact of all markers in all cases is shown in online supplementary table S1.

Inhibition of CXCR7 blocks angiogenesis in a mouse glioma model

To strengthen the limited data on the involvement of CXCR7 in glioma angiogenesis, we tested the effect of CXCR7 inhibition in a subcutaneous glioma mouse model and found that application of CCX771 leads to a significant decrease of microvascular density and inhibited tumour growth (figures 3 and 4), suggesting that, as in other cancer entities,15 CXCR7 is involved in angiogenesis in glial tumours.

Figure 3

Graphical representation of tumour growth (A) and a box-plot representation (B) of the quantification of the microvascular density. The difference between vehicle and the 10 mg/kg or the 30 mg/kg group is statistically significant (p=0.002, Kruskal–Wallis test). Bar represents ∼150 μm.

Figure 4

Block of CXCR7 leads to a significant decrease of microvascular density in a mouse glioma model. Immunohistochemistry for CD31 shows higher microvascular density in a tumour treated only with vehicle (A) as compared with the one treated with a CXCR7 inhibitor (B).


Here we show that the chemokine receptor CXCR7 may have opposite effects on survival based on its cell type specific localisation and the presence of an IDH1 mutation. The mechanism of this interaction is unclear. It is very unlikely that there is a direct link between these two molecules. Most likely, the presence of IDH1 defines a molecular and functional subtype of human gliomas where SDF1a/CXCR7 axis operates in a particular direction (eg, tumour cell invasion or angiogenesis).

By investigating the outcome-relevant expression of molecules with a well-known cellular function, we gained data suggesting that in a predominantly mesenchymal glioma population (PDPN positive) vascular penetration is a positive prognostic feature, while tumour cell invasion influenced outcome negatively. In a non-overlapping (mutIDH1 positive) glioma population, we found the opposite effect of CXCR7 suggesting that angiogenesis there is a poor prognostic feature.

For decades, anti-angiogenic therapy has been one of the most promising therapeutic strategies in oncology. The basic idea of Judah Folkman was to use anti-angiogenic agents in cancer after the ‘angiogenic switch’ in a setting with increased vascular density.16 Bevacizumab, a highly effective anti-angiogenic agent, was approved by the Food and Drug Administration for treatment of recurrent glioblastoma.17 Emerging evidence, however, shows that hypoxia, as a result of anti-angiogenic treatment, may promote tumour invasion and metastasis in several types of cancers18 including glioma.19 Interestingly, experimental studies showed that vascular-targeting approaches promoting tumour oxygenation might prevent the switch to tumour invasiveness.20 Two recent clinical trials investigating the effect of bevacizumab failed to show a substantial survival benefit in newly diagnosed glioblastomas,21 ,22 suggesting that for achieving satisfactory treatment results selection of patients based on molecular or functional characteristics is needed.

The results presented here might have important clinical consequences. Anti-angiogenic therapy has been one of the most promising therapeutic strategies for high-grade gliomas. Our results, however, suggest that anti-angiogenic therapy might be most suitable for IDH1/2 mutant or WHO grade II gliomas. On the contrary, in mesenchymal cases an anti-invasive therapy should be considered. Notably, the expression of an angiogenic factor like vasCXCR7 was a predictor for better outcome in mesenchymal cases suggesting that, as in other cancer types,20 angiogenesis might have a positive prognostic influence in some gliomas most likely by reducing hypoxia and limiting invasiveness. The recently presented preliminary data from Phase III clinical studies showing that anti-angiogenic therapy does not significantly influence overall survival in high-grade gliomas support our hypothesis.21 ,22

Notably within WHO grades, CXCR7 expression did not significantly influence outcome. This might be explained by a coincidence or by the fact that molecular-based subgrouping defines more sharply functional subgroups in human glioma. We are inclined to believe the latter and suggest that an additional glioma classification, extending throughout WHO grades, taking into account genetic signatures and functional characteristics, would be necessary for identifying patient groups responsive to targeted (eg, anti-angiogenic) therapies.


Here we provide evidence suggesting that angiogenesis has a poor prognostic influence in IDH1 mutant glioma cases suggesting that anti-angiogenic therapy might be most suitable for low-grade/IDH1/2 mutant gliomas. On the contrary, in mesenchymal cases an anti-invasive therapy should be considered.

Take home messages

  • Expression of the chemokine receptor CXCR7 on endothelial cells in IDH1 mutant glioma cases predicted poor outcome.

  • In PDPN expressing gliomas CXCR7 predicts diminished prognosis when present on tumour cells and better prognosis when present on endothelial cells.

  • Our data suggest that, although ubiquitously present, angiogenesis and invasion are outcome-relevant events in specific glioma subgroups, providing a potentially important tool for targeted therapy assignment.


We are grateful to Andreas von Deimling, Pieter Wesseling, Herbert Budka, Mark Penfold (ChemoCentryx) and David Reuss for critical discussions and support, and Gerda Ricken and Milena Moneva for excellent technical assistance.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Abstract in German

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Cheok Soon Lee

  • Contributors PB: neuropathological evaluation, statistical analysis, data analysis, manuscript preparation. SN: neuropathological evaluation. AT: cell culture experiments, western blot. JT: data analysis, manuscript preparation. MG: neuropathological evaluation, supervision, data analysis, manuscript preparation.

  • Funding The research was funded using the institution's own research funds.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.