Blood vessels are the body’s highways that allow blood to transport oxygen, nutrients, hormones and waste products quickly and efficiently around the body. Tumors are made up of particularly active cells and so their growth heavily depends on blood vessels. Indeed, a fundamental hallmark of tumor progression is for nearby blood vessels to form more quickly. Tumor blood vessels also differ in structure from their normal counterparts for reasons that need to be investigated in more detail.

Compounds that block the formation of blood vessels have been developed for treating highly malignant brain tumors called gliomas. However, although many of these compounds show promising effects in preclinical trials, clinical trials on humans have been less successful. Having the ability to image the blood vessels in high detail during preclinical trials would help to reveal how treatments that inhibit blood vessel formation work and how tumors might develop resistance to these drugs. However, studying tumor blood vessels remains a challenge due to technical restrictions: techniques that are able to capture how the vessels change over time are unable to show individual cells in much detail, and vice versa.

Magnetic resonance imaging is a versatile tool that can monitor how the blood vessel system of a tumor changes over time in living animals. On the other hand, ultramicroscopy is able to determine the structure of single cells of a particular type. By combining these techniques, Breckwoldt, Bode et al. have now developed a imaging platform that allows the formation of tumor blood vessels to be precisely mapped in the setting of a preclinical study. It also enables detailed investigations into how the structure of the blood vessels is altered by treatments that aim to inhibit the formation and growth of new vessels.

Using this approach on mice with gliomas, Breckwoldt, Bode et al. demonstrated that drugs that inhibit the formation of the blood vessels that supply tumors also cause the blood vessels to take on a more normal structure. Furthermore, treating the mice with a single inhibitory drug was unable to stop tumor growth, mirroring the situation in humans.

Currently, new inhibitors are being developed, offering the possibility of combined treatments that may be more effective than using a single drug on its own. The imaging platform developed by Breckwoldt, Bode et al. will allow the therapeutic effects obtained by these new treatments to be analyzed in detail during preclinical studies.

Gliomas are highly malignant brain tumors with poor prognosis (Wen and Kesari, 2008). Glioblastoma multiforme (GBM) is characterized by high cellular proliferation rates and a rapid induction of angiogenesis (Wen and Kesari, 2008). Angiogenesis is a hallmark of malignant tumors, and most tumors show an exponential ingrowth of neovessels upon a certain tumor size to accommodate their metabolic needs (Carmeliet and Jain, 2011; Hanahan and Weinberg, 2011). This phenomenon is called 'angiogenic switch' and is characterized by the upregulation of pro-angiogenic molecules like vascular endothelial growth factor (VEGF) or angiopoietin 2 (Ang-2) that induce fast tumor angiogenesis. The angiogenic switch is a central feature of malignant tumors and leads to a transition toward a more invasive and aggressive tumor growth (Bruno et al., 2014). Hence, many efforts have been made to develop antiangiogenic therapies to 'starve' tumors off their resources (Huang et al., 2003). So far, this approach of using anti-VEGF agents like bevacizumab has shown clinical benefit in certain tumor types (Mittal et al., 2014) but, despite significant improvement of progression-free survival, does not prolong overall survival of primary GBM in an unselected patient cohort (Chinot et al., 2014). Similarly, VEGF receptor 2 (VEGF-R2) inhibition, despite strong preclinical data (Batchelor et al., 2007; Winkler et al., 2004), has not been successful in neuro-oncology yet (Batchelor et al., 2013). New methods are currently developed to better identify those patients who might benefit from antiangiogenic therapies: Next to molecular approaches (Sandmann et al., 2015), magnetic resonance (MR) imaging could be used for patient stratification. In fact, accumulating data suggest that vascular normalization that occurs early after the initiation of therapy, might be such a marker which could aid clinical decision making (Emblem et al., 2013; Lu-Emerson et al., 2015). However, improved imaging techniques that faithfully and non-invasively characterize vessel architecture and antiangiogenic treatment effects are needed to facilitate the understanding of biological actions of these therapies, and the development of clinical trials.

We employed a MR imaging approach to monitor tumor vessels at single vessel resolution. The technique is based on T2*-weighted (T2*-w) high-resolution Blood Oxygenation Level Dependent (BOLD) venography (Park et al., 2008) with standard gadolinium (Gd) contrast agents at ultrahigh-field strength (9.4 Tesla, isotropic 80 µm resolution). This visualizes substantially more vascular detail compared to conventional T2*-w imaging. We performed pre- and post-contrast MR scans to define angiogenesis during glioma development in two different glioma models. Also, we assess treatment effects of anti-VEGF or radiation therapy on the vascular compartment. We further mapped the tumor vascularization by correlated, dual-color ultramicroscopy (UM) of cleared, unsectioned brains (Ertürk et al., 2012; Schwarz et al., 2015). Fluorescent labeling of the microvasculature using lectins resulted in complementary 3D MR and UM data sets (dubbed 'MR-UM') of the entire tumor 'macro-' and 'microvasculature', which can be compared side-by-side. Thus, MR-UM bridges the gap between MR and optical imaging and may serve as a platform to better understand underlying mechanisms of antiangiogenic treatment and to identify novel druggable targets for preclinical therapy development.

In this study, we combined in vivo 9.4 Tesla MRI with ex vivo UM and tissue clearing to dynamically resolve tumor arterioles and venules as well as the capillary bed of mouse gliomas at high resolution. Contrast-enhanced T2*-w imaging allowed the visualization of neovessels in the 50–100 µm range. In the GL261 model, angiogenesis started 2 weeks post-tumor implantation and massively increased until week 3. One limitation of MRI concerns its resolution, which was limited to ~80 µm in our study. To overcome this limitation, we employed correlated 3D UM of cleared specimen. For UM, we applied iv perfusion with fluorescent lectins that bind specifically to endothelial glycoproteins (Wälchli et al., 2015) and subsequent optical clearing of whole adult mouse brains using the well-established 3DISCO and the recently published FluoClearBABB protocol (Ertürk et al., 2012; Schwarz et al., 2015). We obtained high-quality images with both clearing methods that were used to reconstruct the entire tumor anatomy and blood vessel architecture including bulk mass and invasive zone. Tissue clearing and UM has not been employed for the study of neurooncological disease. Previous UM studies investigated microvessels under physiological conditions (Jährling et al., 2009) and in the context of solid visceral tumors (Dobosz et al., 2014). We have extended this work to the field of neurooncology and combined it with advanced MRI techniques. This bridges the gap between MRI and whole brain optical microscopy, which represent separated domains so far. Optical clearing methods have recently undergone a rapid development and evoked large interest in the neuroscience community as they promise to allow 'connectomic' and circuit reconstruction studies (Chung et al., 2013; Dodt et al., 2007; Ertürk et al., 2011; Keller and Dodt, 2012; Schwarz et al., 2015; Susaki et al., 2014). Clearing techniques are, however, also amenable to the study of pathological states: First reports have highlighted how 3D information obtained by tissue clearing can give new insights into pathophysiology of neuroinflammatory and neurodegenerative disease (Jährling et al., 2015; Spence et al., 2014).

The combination with MR, the most widely used clinical imaging technique, offers added value by correlating spatiotemporal information obtained by MR with high-resolution optical imaging. In the preclinical arena MR and UM with tissue clearing appear as natural partners as both produce 3D datasets, which can be compared side-by-side. The advantages of both techniques can be combined: Namely, MR is a versatile tool for longitudinal in vivo studies. UM of cleared specimen can resolve cellular and subcellular processes and utilize the vast toolbox of genetic and chemical fluorescent labels with their high molecular specificity, thus complementing the information gained by MRI. Also, a combination of the two techniques allows cross-correlation of in vivo signals with molecularly defined optical methods. To show the additive power of both methods, we used a neovascularization paradigm since neovessels are hallmarks of many diseases including cancer (Carmeliet and Jain, 2011; Goveia et al., 2014). Using cross-correlated MR-UM we achieved single vessel resolution and found pathological vessel permeability, tortuosity, and calibers, all of which could be partially reversed by VEGF treatment.

Our data indicate that similar to the human situation, mono-modality antiangiogenic treatment is not sufficient to halt glioma growth. Thus, novel antiangiogenic treatment regimes, possibly in combination with chemotherapy in recurrent disease (as done with promising results in the Dutch Belob Trial (Taal et al., 2014), and the ongoing EORTC 26,101 and 26,091 trials), or in combination with additional treatment modalities (i.e.immunotherapy, chemotherapy) are warranted (Chinot et al., 2014; Kamoun et al., 2009). Anti-angiogenic treatment development and clinical trials require imaging biomarkers to assess vascularization status and treatment effects. The developed T2*-w sequence is suitable to map tumor arterioles and venules in a preclinical setting. Future studies at high-field clinical MR systems should address a possible translation of our MRI approach to the clinical arena.

One limitation of our study relates to the differentiation of the various hierarchical parts of the vascular system. Tumors harbor a manifold of different vascular compartments (veins and arteries of different calibers and sizes with highly variable, but mostly pathological perfusion). Neither MR nor optical methods are able to capture the entire complexity of these compartments. Also, we found that in the bulk tumor some CD31 positive vessels were negative for lectin-FITC, indicating that some tumor vessels were hypo- or non-perfused. This implies that the visualization and quantification of the microvasculature by lectin-injection might slightly underestimate the vascular tumor compartment. The fact that hypo- or non-perfused vessels with abnormal flow exist in tumors is well established (Carmeliet and Jain, 2011). Another drawback might be that clearing methods may lead to an alteration in brain size, that is, due to dehydration. For 3DISCO, we quantified the apparent reduction of brain volume using CT imaging before and after clearing. This revealed a volume reduction of ~40%, which however occurred in a uniform fashion and did not change macroscopic tissue proportions. Anatomical structures and substructures of the brain remained fully intact. Vessel diameters obtained by UM could therefore be corrected by a multiplication factor of ~1.6. However, microscopic shrinkage may slightly differ from the macroscopic factor obtained by CT.

Also, when using endogenous fluorescent proteins such as GFP or RFP in combination with exogenous fluorescent markers such as FITC, preservation of fluorescent signals during the clearing process might vary and should be carefully controlled for.

In summary, the present study shows the additive power of correlative MR and UM for assessing the vascular system and the dynamic changes that occur in tumors over time. The applied principles should be easily transferable to additional targets like the immune cell compartment that are amenable to MR and optical labeling strategies (Kircher et al., 2003; Weissleder et al., 2014). Furthermore, MR-UM should be useful for translational approaches to aid preclinical therapy development. In our study, mono-antiangiogenic treatment with a murine VEGF inhibitor was insufficient to halt tumor growth which mirrors current human studies (Chinot et al., 2014). Dual angiogenesis inhibitors are currently being developed and therapeutic effects could be assessed in detail using our toolkit. Thus, we believe that MR-UM can provide a versatile platform for translational research approaches to cross-correlate MR and optical signals.

1. 1. Batchelor TT
   2. Mulholland P
   3. Neyns B
   4. Nabors LB
   5. Campone M
   6. Wick A
   7. Mason W
   8. Mikkelsen T
   9. Phuphanich S
   10. Ashby LS
   11. DeGroot J
   12. Gattamaneni R
   13. Cher L
   14. Rosenthal M
   15. Payer F
   16. Jurgensmeier JM
   17. Jain RK
   18. Sorensen AG
   19. Xu J
   20. Liu Q
   21. van den Bent M

   (2013) Phase III randomized trial comparing the efficacy of cediranib as monotherapy, and in combination with lomustine, versus lomustine alone in patients with recurrent glioblastoma

   Journal of Clinical Oncology 31:3212–3218.

   https://doi.org/10.1200/JCO.2012.47.2464

   • Google Scholar
2. 1. Batchelor TT
   2. Sorensen AG
   3. di Tomaso E
   4. Zhang W-T
   5. Duda DG
   6. Cohen KS
   7. Kozak KR
   8. Cahill DP
   9. Chen P-J
   10. Zhu M
   11. Ancukiewicz M
   12. Mrugala MM
   13. Plotkin S
   14. Drappatz J
   15. Louis DN
   16. Ivy P
   17. Scadden DT
   18. Benner T
   19. Loeffler JS
   20. Wen PY
   21. Jain RK

   (2007) AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, normalizes tumor vasculature and alleviates edema in glioblastoma patients

   Cancer Cell 11:83–95.

   https://doi.org/10.1016/j.ccr.2006.11.021

   • Google Scholar
3. 1. Bruno A
   2. Pagani A
   3. Pulze L
   4. Albini A
   5. Dallaglio K
   6. Noonan DM
   7. Mortara L

   (2014) Orchestration of angiogenesis by immune cells

   Frontiers in Oncology 4:131.

   https://doi.org/10.3389/fonc.2014.00131

   • Google Scholar
4. 1. Carmeliet P
   2. Jain RK

   (2011) Molecular mechanisms and clinical applications of angiogenesis

   Nature 473:298–307.

   https://doi.org/10.1038/nature10144

   • Google Scholar
5. 1. Chinot OL
   2. Wick W
   3. Mason W
   4. Henriksson R
   5. Saran F
   6. Nishikawa R
   7. Carpentier AF
   8. Hoang-Xuan K
   9. Kavan P
   10. Cernea D
   11. Brandes AA
   12. Hilton M
   13. Abrey L
   14. Cloughesy T

   (2014) Bevacizumab plus radiotherapy–temozolomide for newly diagnosed glioblastoma

   New England Journal of Medicine 370:709–722.

   https://doi.org/10.1056/NEJMoa1308345

   • Google Scholar
6. 1. Chung K
   2. Wallace J
   3. Kim S-Y
   4. Kalyanasundaram S
   5. Andalman AS
   6. Davidson TJ
   7. Mirzabekov JJ
   8. Zalocusky KA
   9. Mattis J
   10. Denisin AK
   11. Pak S
   12. Bernstein H
   13. Ramakrishnan C
   14. Grosenick L
   15. Gradinaru V
   16. Deisseroth K

   (2013) Structural and molecular interrogation of intact biological systems

   Nature 497:332–337.

   https://doi.org/10.1038/nature12107

   • Google Scholar
7. 1. Dobosz M
   2. Ntziachristos V
   3. Scheuer W
   4. Strobel S

   (2014) Multispectral fluorescence ultramicroscopy: three-dimensional visualization and automatic quantification of tumor morphology, drug penetration, and antiangiogenic treatment response

   Neoplasia 16:1–W7.

   https://doi.org/10.1593/neo.131848

   • Google Scholar
8. 1. Dodt H-U
   2. Leischner U
   3. Schierloh A
   4. Jährling N
   5. Mauch CP
   6. Deininger K
   7. Deussing JM
   8. Eder M
   9. Zieglgänsberger W
   10. Becker K

   (2007) Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain

   Nature Methods 4:331–336.

   https://doi.org/10.1038/nmeth1036

   • Google Scholar
9. 1. Emblem KE
   2. Mouridsen K
   3. Bjornerud A
   4. Farrar CT
   5. Jennings D
   6. Borra RJH
   7. Wen PY
   8. Ivy P
   9. Batchelor TT
   10. Rosen BR
   11. Jain RK
   12. Sorensen AG

   (2013) Vessel architectural imaging identifies cancer patient responders to anti-angiogenic therapy

   Nature Medicine 19:1178–1183.

   https://doi.org/10.1038/nm.3289

   • Google Scholar
10. 1. Ertürk A
    2. Becker K
    3. Jährling N
    4. Mauch CP
    5. Hojer CD
    6. Egen JG
    7. Hellal F
    8. Bradke F
    9. Sheng M
    10. Dodt H-U

    (2012) Three-dimensional imaging of solvent-cleared organs using 3DISCO

    Nature Protocols 7:1983–1995.

    https://doi.org/10.1038/nprot.2012.119

    • Google Scholar
11. 1. Ertürk A
    2. Mauch CP
    3. Hellal F
    4. Förstner F
    5. Keck T
    6. Becker K
    7. Jährling N
    8. Steffens H
    9. Richter M
    10. Hübener M
    11. Kramer E
    12. Kirchhoff F
    13. Dodt HU
    14. Bradke F

    (2011) Three-dimensional imaging of the unsectioned adult spinal cord to assess axon regeneration and glial responses after injury

    Nature Medicine 18:166–171.

    https://doi.org/10.1038/nm.2600

    • Google Scholar
12. 1. Goveia J
    2. Stapor P
    3. Carmeliet P

    (2014) Principles of targeting endothelial cell metabolism to treat angiogenesis and endothelial cell dysfunction in disease

    EMBO Molecular Medicine 6:1105–1120.

    https://doi.org/10.15252/emmm.201404156

    • Google Scholar
13. 1. Hanahan D
    2. Weinberg RA

    (2011) Hallmarks of cancer: the next generation

    Cell 144:646–674.

    https://doi.org/10.1016/j.cell.2011.02.013

    • Google Scholar
14. 1. Huang J
    2. Frischer JS
    3. Serur A
    4. Kadenhe A
    5. Yokoi A
    6. McCrudden KW
    7. New T
    8. O'Toole K
    9. Zabski S
    10. Rudge JS
    11. Holash J
    12. Yancopoulos GD
    13. Yamashiro DJ
    14. Kandel JJ

    (2003) Regression of established tumors and metastases by potent vascular endothelial growth factor blockade

    Proceedings of the National Academy of Sciences of the United States of America 100:7785–7790.

    https://doi.org/10.1073/pnas.1432908100

    • Google Scholar
15. 1. Jährling N
    2. Becker K
    3. Dodt H-U

    (2009) 3D-reconstruction of blood vessels by ultramicroscopy

    Organogenesis 5:227–230.

    https://doi.org/10.4161/org.5.4.10403

    • Google Scholar
16. 1. Jährling N
    2. Becker K
    3. Wegenast-Braun BM
    4. Grathwohl SA
    5. Jucker M
    6. Dodt H-U
    7. Vitorica J

    (2015) Cerebral β-amyloidosis in mice investigated by ultramicroscopy

    PLOS One 10:e0125418.

    https://doi.org/10.1371/journal.pone.0125418

    • Google Scholar
17. 1. Kamoun WS
    2. Ley CD
    3. Farrar CT
    4. Duyverman AM
    5. Lahdenranta J
    6. Lacorre DA
    7. Batchelor TT
    8. di Tomaso E
    9. Duda DG
    10. Munn LL
    11. Fukumura D
    12. Sorensen AG
    13. Jain RK

    (2009) Edema control by cediranib, a vascular endothelial growth factor receptor-targeted kinase inhibitor, prolongs survival despite persistent brain tumor growth in mice

    Journal of Clinical Oncology 27:2542–2552.

    https://doi.org/10.1200/JCO.2008.19.9356

    • Google Scholar
18. 1. Keller PJ
    2. Dodt H-U

    (2012) Light sheet microscopy of living or cleared specimens

    Current Opinion in Neurobiology 22:138–143.

    https://doi.org/10.1016/j.conb.2011.08.003

    • Google Scholar
19. 1. Kircher MF
    2. Mahmood U
    3. King RS
    4. Weissleder R
    5. Josephson L

    (2003)

    A multimodal nanoparticle for preoperative magnetic resonance imaging and intraoperative optical brain tumor delineation

    Cancer Research 63:8122–8125.

    • Google Scholar
20. 1. Lemke D
    2. Pfenning P-N
    3. Sahm F
    4. Klein A-C
    5. Kempf T
    6. Warnken U
    7. Schnolzer M
    8. Tudoran R
    9. Weller M
    10. Platten M
    11. Wick W

    (2012) Costimulatory protein 4IgB7H3 drives the malignant phenotype of glioblastoma by mediating immune escape and invasiveness

    Clinical Cancer Research 18:105–117.

    https://doi.org/10.1158/1078-0432.CCR-11-0880

    • Google Scholar
21. 1. Lu-Emerson C
    2. Duda DG
    3. Emblem KE
    4. Taylor JW
    5. Gerstner ER
    6. Loeffler JS
    7. Batchelor TT
    8. Jain RK

    (2015) Lessons from anti-vascular endothelial growth factor and anti-vascular endothelial growth factor receptor trials in patients with glioblastoma

    Journal of Clinical Oncology 33:1197–1213.

    https://doi.org/10.1200/JCO.2014.55.9575

    • Google Scholar
22. 1. Mittal K
    2. Koon H
    3. Elson P
    4. Triozzi P
    5. Dowlati A
    6. Chen H
    7. Borden EC
    8. Rini BI

    (2014) Dual VEGF/VEGFR inhibition in advanced solid malignancies

    Cancer Biology & Therapy 15:975–981.

    https://doi.org/10.4161/cbt.29187

    • Google Scholar
23. 1. Osswald M
    2. Jung E
    3. Sahm F
    4. Solecki G
    5. Venkataramani V
    6. Blaes J
    7. Weil S
    8. Horstmann H
    9. Wiestler B
    10. Syed M
    11. Huang L
    12. Ratliff M
    13. Karimian Jazi K
    14. Kurz FT
    15. Schmenger T
    16. Lemke D
    17. Gömmel M
    18. Pauli M
    19. Liao Y
    20. Häring P
    21. Pusch S
    22. Herl V
    23. Steinhäuser C
    24. Krunic D
    25. Jarahian M
    26. Miletic H
    27. Berghoff AS
    28. Griesbeck O
    29. Kalamakis G
    30. Garaschuk O
    31. Preusser M
    32. Weiss S
    33. Liu H
    34. Heiland S
    35. Platten M
    36. Huber PE
    37. Kuner T
    38. von Deimling A
    39. Wick W
    40. Winkler F

    (2015) Brain tumour cells interconnect to a functional and resistant network

    Nature 528:93–98.

    https://doi.org/10.1038/nature16071

    • Google Scholar
24. 1. Park S-H
    2. Masamoto K
    3. Hendrich K
    4. Kanno I
    5. Kim S-G

    (2008) Imaging brain vasculature with BOLD microscopy: MR detection limits determined by in vivo two-photon microscopy

    Magnetic Resonance in Medicine 59:855–865.

    https://doi.org/10.1002/mrm.21573

    • Google Scholar
25. 1. Sandmann T
    2. Bourgon R
    3. Garcia J
    4. Li C
    5. Cloughesy T
    6. Chinot OL
    7. Wick W
    8. Nishikawa R
    9. Mason W
    10. Henriksson R
    11. Saran F
    12. Lai A
    13. Moore N
    14. Kharbanda S
    15. Peale F
    16. Hegde P
    17. Abrey LE
    18. Phillips HS
    19. Bais C

    (2015) Patients with proneural glioblastoma may derive overall survival benefit from the addition of bevacizumab to first-line radiotherapy and temozolomide: retrospective analysis of the AVAglio trial

    Journal of Clinical Oncology 33:2735–2744.

    https://doi.org/10.1200/JCO.2015.61.5005

    • Google Scholar
26. 1. Sato Y
    2. Nakajima S
    3. Shiraga N
    4. Atsumi H
    5. Yoshida S
    6. Koller T
    7. Gerig G
    8. Kikinis R

    (1998) Three-dimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images

    Medical Image Analysis 2:143–168.

    https://doi.org/10.1016/S1361-8415(98)80009-1

    • Google Scholar
27. 1. Schmitt M
    2. Pawlita M

    (2009) High-throughput detection and multiplex identification of cell contaminations

    Nucleic Acids Research 37:e119.

    https://doi.org/10.1093/nar/gkp581

    • Google Scholar
28. 1. Schwarz MK
    2. Scherbarth A
    3. Sprengel R
    4. Engelhardt J
    5. Theer P
    6. Giese G
    7. Merigan WH

    (2015) Fluorescent-protein stabilization and high-resolution imaging of cleared, intact mouse brains

    PLOS One 10:e0124650.

    https://doi.org/10.1371/journal.pone.0124650

    • Google Scholar
29. 1. Spence RD
    2. Kurth F
    3. Itoh N
    4. Mongerson CRL
    5. Wailes SH
    6. Peng MS
    7. MacKenzie-Graham AJ

    (2014) Bringing CLARITY to gray matter atrophy

    NeuroImage 101:625–632.

    https://doi.org/10.1016/j.neuroimage.2014.07.017

    • Google Scholar
30. 1. Srivastava K
    2. Hu J
    3. Korn C
    4. Savant S
    5. Teichert M
    6. Kapel SS
    7. Jugold M
    8. Besemfelder E
    9. Thomas M
    10. Pasparakis M
    11. Augustin HG

    (2014) Postsurgical adjuvant tumor therapy by combining anti-angiopoietin-2 and metronomic chemotherapy limits metastatic growth

    Cancer Cell 26:880–895.

    https://doi.org/10.1016/j.ccell.2014.11.005

    • Google Scholar
31. 1. Susaki EA
    2. Tainaka K
    3. Perrin D
    4. Kishino F
    5. Tawara T
    6. Watanabe TM
    7. Yokoyama C
    8. Onoe H
    9. Eguchi M
    10. Yamaguchi S
    11. Abe T
    12. Kiyonari H
    13. Shimizu Y
    14. Miyawaki A
    15. Yokota H
    16. Ueda HR

    (2014) Whole-brain imaging with single-cell resolution using chemical cocktails and computational analysis

    Cell 157:726–739.

    https://doi.org/10.1016/j.cell.2014.03.042

    • Google Scholar
32. 1. Taal W
    2. Oosterkamp HM
    3. Walenkamp AME
    4. Dubbink HJ
    5. Beerepoot LV
    6. Hanse MCJ
    7. Buter J
    8. Honkoop AH
    9. Boerman D
    10. de Vos FYF
    11. Dinjens WNM
    12. Enting RH
    13. Taphoorn MJB
    14. van den Berkmortel FWPJ
    15. Jansen RLH
    16. Brandsma D
    17. Bromberg JEC
    18. van Heuvel I
    19. Vernhout RM
    20. van der Holt B
    21. van den Bent MJ

    (2014) Single-agent bevacizumab or lomustine versus a combination of bevacizumab plus lomustine in patients with recurrent glioblastoma (BELOB trial): a randomised controlled phase 2 trial

    The Lancet Oncology 15:943–953.

    https://doi.org/10.1016/S1470-2045(14)70314-6

    • Google Scholar
33. 1. Tofts PS
    2. Kermode AG

    (1991) Measurement of the blood-brain barrier permeability and leakage space using dynamic MR imaging. 1. fundamental concepts

    Magnetic Resonance in Medicine 17:357–367.

    https://doi.org/10.1002/mrm.1910170208

    • Google Scholar
34. 1. Wälchli T
    2. Mateos JM
    3. Weinman O
    4. Babic D
    5. Regli L
    6. Hoerstrup SP
    7. Gerhardt H
    8. Schwab ME
    9. Vogel J

    (2015) Quantitative assessment of angiogenesis, perfused blood vessels and endothelial tip cells in the postnatal mouse brain

    Nature Protocols 10:53–74.

    https://doi.org/10.1038/nprot.2015.002

    • Google Scholar
35. 1. Weissleder R
    2. Nahrendorf M
    3. Pittet MJ

    (2014) Imaging macrophages with nanoparticles

    Nature Materials 13:125–138.

    https://doi.org/10.1038/nmat3780

    • Google Scholar
36. 1. Wen PY
    2. Kesari S

    (2008) Malignant gliomas in adults

    New England Journal of Medicine 359:492–507.

    https://doi.org/10.1056/NEJMra0708126

    • Google Scholar
37. 1. Winkler F
    2. Kozin SV
    3. Tong RT
    4. Chae S-S
    5. Booth MF
    6. Garkavtsev I
    7. Xu L
    8. Hicklin DJ
    9. Fukumura D
    10. di Tomaso E
    11. Munn LL
    12. Jain RK

    (2004) Kinetics of vascular normalization by VEGFR2 blockade governs brain tumor response to radiation

    Cancer Cell 6:553–563.

    https://doi.org/10.1016/j.ccr.2004.10.011

    • Google Scholar

Thank you for submitting your work entitled "Correlated MR imaging and ultramicroscopy (MR-UM) is a tool kit to assess the dynamics of glioma angiogenesis" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

The following individuals involved in review of your submission have agreed to reveal their identity: Pascal Sati and Hans-Ulrich Dodt (peer reviewers). A further reviewer remains anonymous.

Summary:

Breckwoldt et al. investigate angiogenesis in a mouse glioma model. They clearly show that the combination of MRI and ultramicroscopy can provide both longtime observation and high resolution of the tumor as well as the effect of VEGF treatment and photon irradiation.

Essential revisions:

1) Methodology: a major concern is the use of dye injection to assess tumor vascularization. The use of contrast agents is of course justifiable for MR, but to correlate in vivo data with ex vivo structural quantifications, authors should rule out the possible involvement of perfusion in the diffusion of the injected dyes. For a thorough anatomical correlation in the same samples, authors should use well-characterized protocols of whole-mount immunohistochemistry (IHC) using specific vascular markers (e.g. CD31, such as used by iDISCO), which would ascertain that the whole vasculature is visualized. The use of IHC would also allow better staining of smaller capillary beds in which perfusion might have been affected by the presence of the tumor. One option would be to validate, in a subset of samples, the structural data with anti-CD31 IHC. Finally, vascular permeability should be assessed/validated using tracers and conventional methods (e.g., conjugated dextrans, Evans blue) such that, as for tumor angiogenesis and vascular morphology, authors could correlate in vivo and ex vivo data.

2) Interpretations: In the manuscript, one can read terms such as "high-resolution quantification", "single vessel resolution", "ultramicroscopy", or "visualization of venules". However, according to MR data, such resolution is not reached. Larger vessels can indeed be observed at the brain surface under MR, but the resolution is pretty poor (e.g., magnifications in Figure 1b-d, Figure 1i, and Figure 2c,d). Moreover, from panels in Figure 1d, one can ask whether the strong signals are due to presence of vessels or to tracer leakage. In addition, it is dangerous to talk about "neovessels" and "massive angiogenesis" from MR data in Figure 1. Knowing that this MR intrinsically lacks resolution, the reader is hardly convinced that these signals are actually blood vessels. The fact MR relies on perfusion, it seems difficult to decipher between newly formed vessels or increased perfusion in the tumor area. It is only from Figure 2, in correlative experiments with cleared brains, that the reader starts to observe increased vascularization in tumor. Moreover, strong terms such as "increased caliber and tortuousness" are used to describe Figure 1i but none of these features appear in the panels. Finally, it is puzzling why authors chose to analyze "single image planes" from light sheet microscopy in order to quantify vascular morphology. The power of this imaging technique should be leveraged to perform 3D analysis and obtain more detailed measures within tissue volumes (effects would probably reach significance if analyzed in 3D).

3) Translational aspect: Anti-angiogenic therapy to target tumor growth is a classic approach and there is a need to develop better tools to investigate tumor vascularization. However, since apparently the anti-VEGF treatment did not reduce tumor size (as claimed in the text, but unfortunately not illustrated), the rationale for this translational approach is unclear. The authors also mention: "Despite the efficient blockade of neoangiogenesis, anti-VEGF therapy did not lead to reduced tumor growth (…), reflecting current clinical data". This is very vague and one can wonder where these data come from (Method? Number of animals? Graphs? Ref?). Maybe authors can offer an interpretation of why it is failed using their vessel imaging data?

4) An additional technical detail we would mention is that the problem of fluorescence preservation is only severe for endogenous fluorescent proteins like GFP, not really for exogenous fluorescent markers that are much more stable.

1) Methodology: a major concern is the use of dye injection to assess tumor vascularization. The use of contrast agents is of course justifiable for MR, but to correlate in vivo data with ex vivo structural quantifications, authors should rule out the possible involvement of perfusion in the diffusion of the injected dyes. For a thorough anatomical correlation in the same samples, authors should use well-characterized protocols of whole-mount immunohistochemistry (IHC) using specific vascular markers (e.g. CD31, such as used by iDISCO), which would ascertain that the whole vasculature is visualized. The use of IHC would also allow better staining of smaller capillary beds in which perfusion might have been affected by the presence of the tumor. One option would be to validate, in a subset of samples, the structural data with anti-CD31 IHC.

As suggested by the reviewers we have performed immunohistochemistry for CD31 in tissue sections from lectin-injected animals. This resulted in a perfect co-localization of both stainings outside of the tumor mass. In the tumor, however, there were a number of vessels that stained only positive for CD31 but had no apparent lectin staining, indicating non-perfused vessels that were not reached by the lectin injection. We calculated that 49.3% of vessels were double positive for lectin and CD31 within the tumor (179/363 tumor vessels, n=3 mice). 50.7% were only positive for CD31 but not for lectin. The fact that hypo- or non-perfused vessels with abnormal flow exist in tumors is well established (Carmeliet and Jain, 2011). Our utilized method using injected lectins assesses perfused blood vessels and does not account for all tumor vessels. We have added this information to the Results and Discussion and include immunohistochemistry data in the new Figure 3–figure supplement c.

Finally, vascular permeability should be assessed/validated using tracers and conventional methods (e.g., conjugated dextrans, Evans blue) such that, as for tumor angiogenesis and vascular morphology, authors could correlate in vivo and ex vivo data.

We have performed correlated in vivo and ex vivo imaging to assess permeability using Gd and Evans blue (EB) (n=2 mice). We find a good agreement of both tracers that show the increased permeability in the tumor area (correlation coefficient: 0.87, p<0.001). Example images are shown in the new Figure 2–figure supplement 2a-c. One concern would be certainly if there was major leakage of lectin into the extracellular space due to blood-brain-barrier disruption (BBB-D). This however, we don’t see in our model, probably because lectins are much bigger than either gadodiamide or EB and do not passively cross the compromised BBB.

2) Interpretations: In the manuscript, one can read terms such as "high-resolution quantification", "single vessel resolution", "ultramicroscopy", or "visualization of venules". However, according to MR data, such resolution is not reached. Larger vessels can indeed be observed at the brain surface under MR, but the resolution is pretty poor (e.g., magnifications in Figure 1b-d, Figure 1i, and Figure 2c,d). Moreover, from panels in Figure 1d, one can ask whether the strong signals are due to presence of vessels or to tracer leakage. In addition, it is dangerous to talk about "neovessels" and "massive angiogenesis" from MR data in Figure 1. Knowing that this MR intrinsically lacks resolution, the reader is hardly convinced that these signals are actually blood vessels. The fact MR relies on perfusion, it seems difficult to decipher between newly formed vessels or increased perfusion in the tumor area. It is only from Figure 2, in correlative experiments with cleared brains, that the reader starts to observe increased vascularization in tumor. Moreover, strong terms such as "increased caliber and tortuousness" are used to describe Figure 1i but none of these features appear in the panels.

We agree with the reviewers that MRI has intrinsic limitations in terms of resolution (like any other imaging technique). The judgment to call the hypointense structures in Figure 1b-d vessels is based on the fact that most of these hypointense structures are only seen after intravenous contrast administration. We now provide subtraction images of pre and post T2* scans of tumor animals to show the vessel outline that occurs after contrast administration (Author response image 1a). The concept of blood pool imaging by MR contrast agents is not new (Dosa et al., 2011) but has not been widely exploited in the context of tumor vasculature imaging. We agree with the reviewer that MR images need to be interpreted with caution as susceptibility signals are not specific to the vasculature but can be also caused e.g. by bleeding or calcification. Therefore, careful controls and pre- and post-contrast imaging are necessary for interpretation. In our opinion, however, the term “neovessels” and “increased vascularity” seems justified also based on MRI. Our careful longitudinal study allows us to denominate newly appearing hypointensities as neovessels because such structures only appear after contrast administration and can be blocked by VEGF inhibition (longitudinal imaging of one animal, Author response image 1b).

Author response image 1

Download asset Open asset

The term “neoangiogenesis” is mostly referring to the microvasculature (5-10 µm vessels) which we cannot visualize directly by MRI. The vascular signals on MRI most likely represent dilated tumor arterioles and venules that are a result of neoangiogenesis but are not strictly neoangiogenesis itself. We have therefore eliminated the term “neoangiogenesis” from the MRI result section. Permeability and leakage of contrast material into the tissue is certainly a concern. We believe, however, that the linear, vascular-like patterns of the post contrast images (e.g. Figure 1d) show the tumor blood pool and not mere leakage (as this should result in a uniform signal decrease within the entire tumor). Nevertheless, we have taken care to not “over-interpret” our MR data and have eliminated interpretations from the Result section. We now refer to vascular patterns mostly in the ultramicroscopy section and the Discussion.

Finally, it is puzzling why authors chose to analyze "single image planes" from light sheet microscopy in order to quantify vascular morphology. The power of this imaging technique should be leveraged to perform 3D analysis and obtain more detailed measures within tissue volumes (effects would probably reach significance if analyzed in 3D).

We now perform 3D analysis of entire ultramicroscopy data stacks (100-200 single planes) using Amira software. The vasculature is segmented in 3D and vessel radii and pathlength are quantified. Both parameters show highly significant differences between tumor and healthy brain. This data is now included in the extended Figure 3e-h and described in the methods and Results section.

3) Translational aspect: Anti-angiogenic therapy to target tumor growth is a classic approach and there is a need to develop better tools to investigate tumor vascularization. However, since apparently the anti-VEGF treatment did not reduce tumor size (as claimed in the text, but unfortunately not illustrated), the rationale for this translational approach is unclear. The authors also mention: "Despite the efficient blockade of neoangiogenesis, anti-VEGF therapy did not lead to reduced tumor growth (…), reflecting current clinical data". This is very vague and one can wonder where these data come from (Method? Number of animals? Graphs? Ref?). Maybe authors can offer an interpretation of why it is failed using their vessel imaging data?

As described in Figure 4a,b we performed VEGF or isotype control treatment, monitored by MRI (4 mice per group, i.p. treatment with 10 mg/kg VEGF inhibitor or IgG control, every three days). The tumor size increased significantly after one week of treatment (p=0.02) and was not significantly reduced by VEGF treatment compared to the isotype control group (mean tumor size before treatment: 0.076 cm2; after VEGF inhibitor treatment: 0.22 cm2; after isotype treatment: 0.22 cm2; p=n.s., Author response image 1c). Along these lines two recent clinical trials did not show an increased overall survival for bevacizumab treatment compared to standard radio-chemotherapy treatment for primary or recurrent glioblastoma (Chinot et al., 2014; Taal et al., 2014). This information is now included in the Results section. There has been speculation that inhibition of angiogenesis and vascular normalization leads to an activation of tumor stem cells with potential deleterious effects (Lathia et al., 2015). Also our ultramicroscopy data showed only partial vascular normalization and VEGF inhibition did not completely prevent vessel caliber increases and the formation of pathological vessels. This might explain the persistent tumor growth, which was not halted by VEGF inhibition.

4) An additional technical detail we would mention is that the problem of fluorescence preservation is only severe for endogenous fluorescent proteins like GFP, not really for exogenous fluorescent markers that are much more stable.

In this manuscript we use td-tomato as fluorescent protein and FITC/Texas-red as fluorescent dyes and both are preserved by our clearing protocols. However we have added the following cautionary note to the Discussion: “When using endogenous fluorescent proteins such as GFP or RFP in combination with exogenous fluorescent markers such as FITC, preservation of fluorescent signals might vary during clearing and should be carefully controlled for.”

Author details

1. Michael O Breckwoldt

   1. Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany
   2. Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany

   Contribution

   MOB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Julia Bode, Martin Bendszus and Björn Tews

   For correspondence

   michael.breckwoldt@med.uni-heidelberg.de

   Competing interests

   The authors declare that no competing interests exist.

2. Julia Bode

   1. Schaller Research Group, University of Heidelberg and German Cancer Research Center, Heidelberg, Germany
   2. Molecular Mechanisms of Tumor Invasion, German Cancer Research Center, Heidelberg, Germany

   Contribution

   JB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Michael O Breckwoldt, Martin Bendszus and Björn Tews

   Competing interests

   The authors declare that no competing interests exist.

3. Felix T Kurz

   Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

   Contribution

   FTK, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

4. Angelika Hoffmann

   Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

   Contribution

   AH, Acquisition of data, Drafting or revising the article, Contributed unpublished essential data or reagents

   Competing interests

   The authors declare that no competing interests exist.

5. Katharina Ochs

   1. Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany
   2. Molecular Mechanisms of Tumor Invasion, German Cancer Research Center, Heidelberg, Germany

   Contribution

   KO, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

6. Martina Ott

   1. Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany
   2. Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany

   Contribution

   MO, Conception and design, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

7. Katrin Deumelandt

   1. Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany
   2. Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany

   Contribution

   KD, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

8. Thomas Krüwel

   Department of Diagnostic and Interventional Radiology, University Medical Center Göttingen, Göttingen, Germany

   Contribution

   TK, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

9. Daniel Schwarz

   Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

   Contribution

   DS, Conception and design, Acquisition of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

10. Manuel Fischer

    Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    MF, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents

    Competing interests

    The authors declare that no competing interests exist.

11. Xavier Helluy

    1. Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany
    2. NeuroImaging Centre, Research Department of Neuroscience, Ruhr-University Bochum, Bochum, Germany

    Contribution

    XH, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents

    Competing interests

    The authors declare that no competing interests exist.

12. David Milford

    Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    DM, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents

    Competing interests

    The authors declare that no competing interests exist.

13. Klara Kirschbaum

    Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    KK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

14. Gergely Solecki

    Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    GS, Acquisition of data, Drafting or revising the article, Contributed unpublished essential data or reagents

    Competing interests

    The authors declare that no competing interests exist.

15. Sara Chiblak

    1. German Cancer Consortium and Heidelberg Institute of Radiation Oncology, National Center for Radiation Research in Oncology, Heidelberg, Germany
    2. Heidelberg University School of Medicine, Heidelberg University, Heidelberg, Germany
    3. Translational Radiation Oncology, German Cancer Research Center, Heidelberg, Germany

    Contribution

    SC, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

16. Amir Abdollahi

    1. German Cancer Consortium and Heidelberg Institute of Radiation Oncology, National Center for Radiation Research in Oncology, Heidelberg, Germany
    2. Heidelberg University School of Medicine, Heidelberg University, Heidelberg, Germany
    3. Translational Radiation Oncology, German Cancer Research Center, Heidelberg, Germany

    Contribution

    AA, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

17. Frank Winkler

    Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    FW, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

18. Wolfgang Wick

    1. Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany
    2. Clinical Cooperation Unit Neurooncology, German Cancer Consortium, German Cancer Research Center, Heidelberg, Germany

    Contribution

    WW, Conception and design, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

19. Michael Platten

    1. Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center, Heidelberg, Germany
    2. Neurology Clinic and National Center for Tumor Diseases, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    MP, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

20. Sabine Heiland

    Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    SH, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

21. Martin Bendszus

    Neuroradiology Department, University Hospital Heidelberg, Heidelberg, Germany

    Contribution

    MB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Contributed equally with

    Michael O Breckwoldt, Julia Bode and Björn Tews

    Competing interests

    The authors declare that no competing interests exist.

22. Björn Tews

    1. Schaller Research Group, University of Heidelberg and German Cancer Research Center, Heidelberg, Germany
    2. Molecular Mechanisms of Tumor Invasion, German Cancer Research Center, Heidelberg, Germany

    Contribution

    BT, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

    Contributed equally with

    Michael O Breckwoldt, Julia Bode and Martin Bendszus

    For correspondence

    b.tews@dkfz-heidelberg.de

    Competing interests

    The authors declare that no competing interests exist.

• 1,639

  Page views

• 403

  Downloads

• 31

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.