The nature and applications of VDAs have been the subject of many earlier reviews 4,38,87-89
and here we provide an update, particularly featuring the relevance of imaging techniques in assessing VDA activity. There is hope and promise that Radiology can move beyond an anatomical approach to provide effective biomarkers of potential response to a therapy and early indication of therapeutic efficacy.90
Tumor vasculature is recognized to be highly disorganized and inefficient. Many investigators have demonstrated the tortuous mesh of microvessels characterized by blind ends, constrictions and loops causing non-linear flow, as elegantly revealed in vascular casts, such as the classic studies of Konerding.91
We show a typical corrosion cast in , achieved by infusing a liquid monomer into a rat breast tumor growing in the kidney of a rat. Once the material had polymerized, tissue was removed by caustic maceration, the cast coated with a sputtering of palladium-gold and scanning electron microscopy (SEM) performed. Many reports show such corrosion casts in some cases encompassing whole animals.92
Such casts provide an indication of vasculature in three dimensions, though quantitative analysis is not trivial potentially requiring micro CT.93-95
The polymer filaments are fragile and sometimes the finest capillaries may be lost. The cast provides no dynamic information and polymer may be forced into vessels otherwise occluded by temporary thromboses.
Vascular heterogeneity in tumors identified ex vivo
More commonly, vascular extent is assessed using histological specimens with immunohistochemistry, e.g
., reveals blood vessels based on anti-CD31 monoclonal antibody binding. Vascular perfusion may also be observed by infusion of a vascular reporter prior to sacrifice. shows distribution of Hoechst 33342 dye, which had been infused intravenously 60 seconds prior to sacrifice. Overlay on the CD31 image reveals the fraction of vessels that were perfused (). We have used this approach to show the change in vascular extent and perfusion in tumors with respect to administration of typical VDAs.96,97
Two hours after administration of combretastatin A-4P vasculature was detected based on CD31 (similar to baseline), but essentially all perfusion had ceased. Of course, such measurements generally require separate specimens (tumors) for each time point relying on the similarity of matched tumor-pairs. Hoechst dye extravagates from the vasculature, whereas other markers may reveal perfusion based on endothelial binding (e.g
., tomato lectins) or trapping of fluorescent or radioactive microspheres.98
Indeed, sequential administration of stains of different colors before and after an intervention in a pulse chase fashion can reveal dynamic changes post facto 99
as demonstrated by Chaplin et al.
with respect to vascular collapse following administration of the vasodilator hydralazine.100
Superficial vasculature may also be assessed in vivo
using intra vital microscopy particularly as applied to window chamber models.101
Vascular development may be examined repeatedly over a period of days and, with respect to drug interventions and video loops, has revealed passage of individual erythrocytes sometimes revealing fluctuations and reversal of flow within individual vessels.102,103
Addition of fluorescent markers of pH and hypoxia has allowed correlation of multiple physiological parameters in small regions at microscopic resolution.104
While microscopy is often limited to a surface depth of about 100 um, multi-photon approaches have revealed three-dimensional structure and multimodality interactions as represented by photoacoustic tomography allow deeper penetration.105
Laser Doppler flowmetry has provided non-invasive evidence for acute response to CA4P.106
Pre-clinical studies allow sacrifice and high resolution post-mortem analysis of tissues, but a requirement for biopsies to define vascular extent is less satisfactory in patients. It is particularly unsatisfactory for evaluating vascular dynamics since samples may not be representative of the whole tumor and the same tissue cannot be examined twice precluding dynamic studies. Thus, there is a need for non-invasive approaches to reliably, reproducibly, and simply examine changes in tumor vasculature at depth. Today, this is most commonly achieved using dynamic contrast enhanced (DCE) proton MRI (magnetic resonance imaging).107
MRI is widely available, measurements are non-invasive and semi-quantitative measurement is facile based simply on changes in tissue water signal in response to distribution of paramagnetic contrast agent. Indeed, DCE MRI is now included in the clinical development of many VDAs.108-111
Dynamic computer tomography (CT) has also been used,112
but it is often considered to be less attractive due to the radiation dose of repeated CT scans and potential for anaphylactic response to iodinated contrast agents or other adverse side effects.112,113
Paramagnetic gadolinium based contrast agents are recognized to be much safer. Recently, a few instances of NSF (nephrogenic systemic fibrosis) have been reported, but these appear to be associated with poor renal function.114
DCE does require a bolus IV (intravenous) infusion of contrast agent and there is extensive discussion of the optimal procedures regarding data acquisition (temporal and spatial resolution) and interpretation (analytical approaches).115-120
Quantitative analysis is generally more relevant to subtle investigations of angiogenesis and vascular leakiness, while the beauty of many VDAs is the massive acute vascular response, which is readily detected even with simple semi-quantitative analyses.
Pre-clinical studies allow much greater versatility in imaging methodology and we show techniques in our laboratory both emulating the successes of other investigators and introducing novel paradigms.
DCE MRI has been most widely applied in the development of the VDAs. Measurements are non-invasive though they do require the IV infusion of a contrast agent. Essentially all imaging approaches require that animals be anaesthetized, but modern fluorinated gaseous anesthetics, such as isoflurane and sevoflurane, appear to be much less vasoactive, toxic and perturbing than earlier agents such as halothane or pentobarbital or ketamine.121
MRI can provide high temporal resolution and may generate 3D data sets for whole tumor coverage. More typically, a single slice through the center of a tumor is examined, since this reveals heterogeneity (e.g
., differential response of tumor center and periphery) with high temporal resolution. Assessment of vascular dynamics requires administration of sequential doses of contrast agent and measurements could be perturbed by wash-out of residual material from prior measurements. This can be overcome by increasing the dose of successive injections or simply allowing a sufficient interval for wash-out (generally, greater than 30 mins.) and most reports have used intervals of two hours or more between examinations. Since effective VDAs generally cause massive acute effects, experimental protocols and interpretation are quite facile. Even if an animal is removed from the magnet, precluding precise correlation of individual voxels, large regions tend to behave similarly and data are readily compared based on histograms or spatial consideration of regions of interest. Animals may be allowed to wake up between scans, but it may generally be assumed that subtle physiological changes attributable to tumor development are minimal over a few hours. Thus, observed changes due to VDAs are readily identified. This is very different from antiangiogenesis agents, which normally act over days, and thus any changes in vasculature must be separated between “normal” tumor progression and response to drug.122
There are extensive reports of DCE MRI applied to many VDAs including combretastatin A-4P (Zybrestat™)96,97,106,123-126
and combretastatin A-1P (Oxi4503),127
5,6-dimethylxanthenone 4-acetic acid (DMXAA, also called ASA404 (vadimezan),128,129
In many cases simple DCE used small paramagnetic contrast agents, but in other cases larger materials designed to be retained in the vasculature such as macromolecular contrast agent albumin-gadolinium diethylenetriaminepentaacetate (albumin-GdDTPA)136
or SPIOS were used. Diverse tumors have been examined for research in animals (mice and rats) and as part of clinical trials in patients.135
Several investigators have taken the opportunity to use MRI to compare the efficacy of different VDAs.124,137,138
Such measurements have accelerated development of agents providing insight into efficiency, dosing, timing and heterogeneity of activity. An example from our laboratory is in for a 13762NF rat breast tumor with respect to a single dose of combretastatin A-4P administered intraperitoneally (IP). At 2 h, vascular perfusion was severely reduced and delayed, but substantial recovery was observed at 24 h, notably in the tumor periphery. We have presented more extensive data in this tumor system and in a mouse tumor previously.96,97
Analyses of DCE MRI with respect to VDAs have used various levels of complexity ranging from changes in relative si gnal intensity following infusion of contrast agent (semi-quantitative) to rigorous calculation of perfusion fraction, vascular leakage and transit times. Ultimately, parameters are required to reflect efficiency and data reduction may provide averaged values such as mean and median or perfused fractions.
Evaluating VDAs non-invasively by MRI
Alternate vascular dependent contrast mechanisms might be exploited including vascular spin labeling, though often tumors have such small blood vessels with sluggish flow that measurements are impractical. Oxygen may be considered as a contrast agent with changes in BOLD (Blood Oxygen Level Dependent) or TOLD (Tissue Oxygen Level Dependent) contrast response accompanying oxygen breathing challenge before and after the VDA administration. Certainly, tumor vascular extent has been correlated with BOLD response139,140
and flow must be considered, as noted in the FLOOD concept.141
Dynamic response to a hyperoxic gas challenge may reveal vascular shut down, but direct response to drug alone may be confused by coincidental changes in local hematocrit (blood volume), fraction of deoxyhemoglobin, and flow and Thomas et al
. reported a complex pattern in response to carbogen challenge following CA4P treatment in rat bladder tumors growing in nude mice.142
Indeed, Howe et al.,143
have observed apparently contradictory results whereby BOLD signal increased following death, attributable to vascular collapse and deoxyhemoglobin clearance rather than improved oxygenation.
In this regard, vascular volume and oxygenation may be monitored directly using near infrared spectroscopy (NIRS) noting the differential absorption coefficients of oxy- and deoxyhemoglobin.144-146
To date, NIRS has generally lacked spatial resolution, but multi-exponential behavior implies heterogeneity.
Increasingly, it is recognized that combined therapy approaches are needed to successfully treat patients, particularly in the case of VDAs, which often leave a surviving peripheral tumor rim causing rapid tumor reoccurrence. Vascular shut down has implications for concomitant chemotherapy based on effective drug delivery and retention. It is also crucial for combination with radiotherapy, where vascular occlusion is expected to cause regional hypoxia, and hence, radio resistance. Indeed, several studies have shown that the combination of irradiation and VDA is crucially dependent on timing.147-148
We recently examined tumor oxygen dynamics directly based on 19
F MRI oximetry with respect to VDA. Using FREDOM (Fluorocarbon Relaxometry Using Echo Planar Imaging for Dynamic Oxygen Mapping)140
we found significant acute hypoxiation in the 13762NF rat breast tumor within 30 min of administering combretastatin A-4P.96
Heterogeneous regional re-oxygenation was observed 24 h later. An example of such a measurement is shown in , although here the hypoxiation was a little slower. Crucially, sequential pO2
measurements are non-invasive and can be repeated every 6½ min. In comparison, DCE approaches require repeated administration of the contrast agent requiring a priori
choice of measurement times. Such pO2
measurements may be accelerated further by using a Look-Locker approach (90 s) as presented recently by Gallez, et al. 149
or based on a partial saturation measurement (1 s in a perfused heart 150
). Most significantly, such measurements allowed us to optimize timing of combined irradiation and combretastatin to enhance tumor growth delay.151
Assessment of hypoxia accompanying vascular disruption
Vascular imaging may also be achieved using ultrasound,152
notably, with the availability of the new small animal VisualSonics systems, which can provide microscopic resolution or exploit micro bubble contrast agents. Doppler approaches are attractive since they require no contrast agent, hence avoiding the associated costs and technical challenge of IV administration. However, sluggish perfusion of small vessels may handicap observations in some tumors. In , we show vascular changes based on Power Doppler in a rat breast tumor, but the effect is quite subtle. In other tumors, we have seen much more extensive vasculature (manuscript in preparation). Vascular shutdown was readily apparent in this tumor based on infusion of contrast micro bubbles (). More extensive ultrasound studies have been reported by others, notably with respect to vascular disrupting agents or vascular flare following irradiation.153-159
As with MRI, such measurements may be applied clinically.
Assessment of acute changes in tumor vasculature using ultrasound
We recently introduced a novel approach exploiting dynamic bioluminescent imaging (dBLI) to investigate the acute effects of vascular disrupting agents.97
Various reports have considered the dynamics of light emission for luciferase expressing cells growing in tumors in animals following the administration of luciferin substrates.160-162
Most reports have focused on magnitude and duration of light emission together with reproducibility, e.g
., intravenous administration gives most rapid and intense, yet highly transient, light emission kinetics, while intraperitoneal administration is technically easier and gives a longer signal plateau, so that the timing of imaging acquisition is less critical.163
However, we and others have noted a substantial failure rate with no (or minimal) light emission being observed. On the other hand, we find that subcutaneous (SC) administration of luciferin in the back/neck region provides highly reproducible light emission kinetics.164
Noting that light emission requires delivery of luciferin substrate to the tumors by the vasculature this provides an effective assay of vascular patency. We have shown that following administration of CA4P to nude mice with human breast tumor xenografts consistent results were achieved using dynamic BLI or dynamic contrast enhanced MRI.97
MRI does of course provide spatial information including potentially 3D representations. An example of dBLI is shown in for human prostate PC3-Luc tumors growing in two nude mice. Each mouse shows intense BLI signal prior to CA4P with diminished signal at 2 h and significant recovery at 24 h. Kinetic curves of light emission for one of the tumors are shown in the graph. We have now applied this procedure to several VDA drugs, disease sites and tumor types. The method is particularly simple to implement, cheap and offers high throughput. The primary drawback of this approach is the need for luciferase expressing cells.
Assessment of acute changes in tumor vasculature using dynamic bioluminescent imaging (dBLI)
In essence, any technology providing signal sensitive to vascular extent and flow may be applied to investigate VDA activity. In other cases, radionuclides have been used in conjunction with autoradiography, biodistribution, PET and SPECT.165-167