PDT of CNV associated with AMD represents a major step forward in
the treatment of this disease, and its optimization is a focal point of
the present investigation. Although BPD-MA has shown efficacy in phase
III trials
9 10 for treating the classic form of the
exudative type of this disease, the ideal photosensitizer for this
disease has not been found yet. Beyond having high selectivity for the
choroidal neovasculature and high photosensitizing activity, it should
be nontoxic, be rapidly eliminated from the body, show small
interpatient variations in pharmacokinetics, and cause minimal skin
photosensitization. Moreover, the drug should absorb in the red or the
near infrared part of the spectrum for better optical transmission
through ocular tissues than visible wavelengths and less scattering in
patients with cataract.
Cell culture experiments are not optimal for obtaining information on
the vascular damage efficacy of photodynamic agents.
32 Thus, new methods should be developed to provide an effective, low-cost
drug-screening system for promising photosensitizers. The CAM
assay
36 has often been used to study vascular and
angiogenic effects in vivo. Although this model has already been used
for PDT purposes,
29 30 31 32 33 most approaches were not
specifically adapted for AMD-PDT. Topical application and IV injection
represent significantly different uptake mechanisms (i.e., the usual
simple deposition of the photodynamic agent on the CAM surface is quite
different from the clinical situation
29 31 32 in which
intravenous injections are used). The same applies to IP
administration.
29
The present study shows a modification of the CAM model to make it
serve as a low-cost, simple, informative, and appropriate assay for the
prediction of photosensitizer characteristics related to PDT-induced
vascular damage encountered in clinical tests. To evaluate the modified
CAM procedure for the assessment of new photoactive drugs in AMD-PDT,
it is essential to draw a comparison to clinical aspects. For this
purpose, among other observations, two important properties in PDT for
CNV in the human eye are suitable. First, the leakage out of the CNV of
the dye after intravenous injection, and second the response to PDT in
closing the CNV.
The time dependence of the fluorescence of the sensitizer inside and
outside the blood vessels is of importance for two principle reasons.
First, fast leakage of small, water-soluble dyes, like fluorescein,
through fenestrated neovessels is frequently used for the diagnosis of
CNV in the human eye.
9 10 Note that this technique has
been used to quantitatively monitor the effect of PDT treatments of
this disease.
37 Second, a photodynamic agent should not
leak out of these fenestrated neovascular structures and
choriocapillaries too quickly, to prevent photodynamic action on
neighboring structures such as the RPE or the
photoreceptors.
19
The results from the fluorescence pharmacokinetics measurements in our
modified CAM model permit observation of the relations between the
leakage of neovessels observed in the wet form of AMD, in which leaking
fluorescein demarcates CNV in the early phase of the fluorescence
angiographies.
Husain et al.
24 have studied the leakage of BPD-MA in
experimental neovascularization induced in the iris of cynomolgus
monkeys. They found only minimal leakage but a significant decrease of
vascular BPD-MA fluorescence intensity within 10 minutes. Although most
of the BPD-MA was associated with blood lipoproteins among which are
LDLs, they found no differences between the fluorescence
pharmacokinetics of neovessels and that of normal vascularization. This
is in agreement with our observations that the fluorescence time
profiles are similar for blood vessels of different diameter.
One critical point in the search of a valid new drug-screening
procedure for AMD-PDT is to establish an at least semiquantitative
relation with clinical results. Thus, in our experimental setup we
deliberately chose conditions that were close to those used in the
clinic for drug administration and damage assessment. To evaluate
vascular damage in our model, we injected BPD-MA, for which
considerable clinical information is available. Using this
photosensitizer, a drug dose of 6 mg/m
2 corresponding to approximately 0.17 mg/kg for an average adult, is
injected. Light doses of 50 J/cm
2 were used,
because higher light doses clinically have closed retinal vessels and
have impaired the patient’s visual acuity.
9 Supposing a
typical diameter of approximately 30 μm of vessels occurring in
CNV,
38 a successful AMD PDT with BPD-MA would be ranked 3
on our arbitrary damage scale. In our model this value was achieved
using a photosensitizer dose of 4 μg/embryo and a light dose of 15
J/cm
2 or a photosensitizer dose of 2 μg/embryo
and a light dose of 40 J/cm
2 (Fig. 6) . Based on
the mean weight estimates for embryos
39 on EDD13 (∼13 g)
this corresponds to photosensitizer doses of 0.3 and 0.15 mg/kg,
respectively. In clinical practice, however, irradiation is performed
at 690 nm, in which absorption of BPD-MA is approximately 2.5 times
smaller than at 440 nm. This should be taken into account in subsequent
animal tests in which other irradiation wavelength may be more
favorable. In contrast, in the CAM, inner filters, such as melanin in
the human RPE, that prevent the light in the blue spectral region from
reaching the deeper choroidal vasculature are absent. Thus, it is
possible to work with excitation light of approximately 440 nm, which
enables use of standard epifluorescence microscopes, providing fluence
rates in these wavelength ranges similar to those used in clinics. In
biologic systems the internal conversion from the S2 to the S1 state is
rapid, compared with processes such as fluorescence or intersystem
crossing. Thus, using 440 nm instead of 690 nm increases the total
number of excited photosensitizer molecules but does not influence the
fluorescence or singlet oxygen quantum yields.
To the best of our knowledge there are at present no data regarding the
stability of liposomal formulated BPD-MA. For this purpose we included
a series of embryos, injected with previously prepared BPD-MA solutions
in our study protocol. Although fluorescence intensities as well as the
fluorescence distribution after IV injection under these conditions
were found to be similar to those of freshly prepared BPD-MA solution,
we observed significantly reduced phototoxicity. Whether this was due
to the instability of the liposomes containing egg
phosphatidylglycerol and phosphatidylcholine
40 41 42 and
thus reduced binding to endothelial structures
43 or to the
degradation of the photosensitizer itself requires additional
investigations.
To assess reproducibility of our model, we used an extraordinary high
number of eggs per data point. However, because of the small variations
observed under our experimental conditions, it is probable that the
total number of eggs can be significantly reduced for a preliminary
evaluation of the efficacy of photosensitizers.
The presented bioassay provides useful information for assessing
vascular damage after photodynamic treatment. Because of the
architecture of the CAM, it is not clear to what extent the present
model addresses the selectivity of new photosensitizers, because
established vessels such as those in human CNV are absent. As mentioned
earlier, however, the CAM model may easily be extended by the
gel–nylon mesh technique
44 or the gelatin sponge
assay,
34 which profits from newly formed blood vessels
perpendicularly grown to the plane of the CAM into the graft, which
does not contain preexisting vessels.
Unfortunately, other structures of interest in the treatment of CNV
associated with AMD such as the RPE, Bruch’s membrane, or the neural
retina are not present in our model. Again, the model is not intended
to completely replace animal models for in vivo testing of
photosensitizer selectivity, but it will help to minimize the number of
time-consuming and expensive animal tests.
Finally, the present use of the CAM model offers the advantage of easy
damage assessment performed by R101-FA 24 hours after PDT. This
supplementary injection of a fluorescing dye enables observation of the
blood flow through irradiated vessels
33 and the collection
of quantitative information on the real-time status of the blood
circulation within the irradiated area, compared with nonirradiated
zones. The recording of digital fluorescence image before, during, and
24 hours after irradiation allows at least semiquantitative evaluation
of vascular PDT effects using standard imaging processing software.
Even, a follow-up longer than 1 day (up to 2–3 days) can be performed,
when necessary. Because both irradiated and nonirradiated areas are
available in one fluorescence image, numerical evaluation can provide
information on PDT effects of the proliferation pattern of the
capillary network, blood vessel development, and other factors related
to angiogenic processes. Furthermore, a direct, quantitative assessment
of photothrombic effect due to PDT is given by numerical comparison
BPD-FA images during PDT.
In conclusion, we have demonstrated that this new technique, consisting
of IV injection and subsequent fluorescence angiography, transforms the
CAM assay into a model that is appropriate for rapid, inexpensive drug
screening of photosensitizers for vascular effects. It has been shown
to correlate well with clinical observations and will be a major aid in
the selection of promising new for PDT of CNV associated with AMD.
Furthermore, it is sufficiently close to clinical reality that it
appears to be able to predict treatment regimens, which will
considerably reduce the number of animals necessary for clinical
testing during the late stages of drug development.
The authors thank Lucienne Juillerat for initial help in
CAM preparation and for access to some of her laboratory equipment and
Michel Sickenberg for many fruitful discussions.