Abstract
purpose. There is controversy about which mode of laser irradiation, early
irradiation with low-dose photosensitizer or late irradiation with
high-dose, benefits the selective occlusion of choroidal
neovascularization (CNV) in photodynamic therapy (PDT). In this study,
using an amphiphilic photosensitizer, 13,17-bis (1-carboxypropionyl)
carbamoylethyl-8-etheny-2-hydroxy-3-hydroxyiminoethylidene-2,7,12,18-tetraethyl
porphyrin sodium (ATX-S10(Na); Photochemical Inc., Okayama, Japan),
photodynamic and adverse effects of early irradiation on CNV-bearing
monkey eyes were investigated.
methods. Experimentally induced CNV lesions and normal retina were irradiated
with a diode laser (670-nm wavelength) at a dose of 1 to 90
J/cm2 at 1 to 19 minutes after intravenous injection of 2
mg/kg body weight of ATX-S10(Na). Vascular occlusion and CNV recurrence
were evaluated by fluorescein and indocyanine green angiography and
histologic analysis, until 4 weeks after irradiation.
results. Of 45 different conditions, 23 did not induce CNV closure, 20 provided
both CNV occlusion and retinal vessel damage, and 2 achieved selective
CNV occlusion without retinal vascular injury. Recurrence of CNV was
induced in 19 of 22 CNV-occluding conditions. ATX-S10(Na) angiography
showed that dyes were similarly distributed between normal vessels and
CNV at early time periods after injection, whereas they were
preferentially accumulated in CNV after 30 minutes.
conclusions. In PDT with ATX-S10(Na), irradiation within 20 minutes of dye injection
failed to induce selective CNV occlusion, probably because there is no
significant difference in the biodistribution of dye between CNV and
retinal vessels. It also caused frequent CNV recurrence after extensive
inflammation in the irradiated retina.
In patients with age-related macular degeneration (AMD),
choroidal neovascularization (CNV) often causes severe visual loss.
Laser photocoagulation has long been used as the therapy for this
pathologic event,
1 but it has shown limited efficacy,
depending on the site of CNV occurrence in the fundus. For example,
when CNV appears in the subfoveal region, this therapy often causes a
sudden visual loss due to thermal coagulation-related injuries in the
sensory retina.
2 Other treatment modalities, such as
radiation therapy
3 and surgical removal of
CNV,
4 do not always give satisfactory results, and,
furthermore, the effectiveness of macular translocation
surgery
5 still remains unclear.
At present, photodynamic therapy (PDT) is the most promising modality
for CNV occlusion. Five photosensitizers are currently under
investigation.
6 7 8 9 10 11 12 13 14 15 16 17 They are categorized into three groups
of dyes: lipophilic, hydrophilic, and amphiphilic. Lipophilic dyes
include benzoporphyrin monoacid (BPD-MA)
8 9 10 11 12 13 14 15 and
tin-ethyl etiopurpurin (SnET2),
16 the former of which is
the only agent approved for clinical use in the treatment of CNV in
AMD,
13 14 15 and the latter of which is undergoing testing
in a phase III clinical trial. Compared with placebo, BPD-MA shows a
significantly higher rate of visual preservation in patients with
predominantly classic CNV, in which classic CNV occupies more than 50%
of the neovascular lesion.
13 14 15 In the clinic, for
intravenous administration, lipophilic photosensitizers are used in the
form of liposomes. Hydrophilic photosensitizers include lutetium
texaphyrin and mono-
l-aspartyl chlorin e6,
17 the former of which is undergoing testing in a phase-I/II clinical
trial and the latter of which is in the preclinical stage. Amphiphilic
dyes that possess both lipophilic and hydrophilic properties include a
novel photosensitizer ATX-S10(Na), which we have recently developed as
a potent agent for PDT.
18 19 20 21
Differences in chemical structures among photosensitizers lead to
differences in subcellular localization in the target cell. Liposomal
BPD-MA and SnET2 are bound to low-density lipoproteins (LDLs) in the
blood stream, taken up by the target cell through LDL receptor-mediated
endocytosis, and diffusely distributed in the cytoplasm.
22 Hydrophilic dyes are taken up by the target cell through endocytosis
and preferentially accumulate in the lysosomes. An amphiphilic dye
ATX-S10(Na) is conjugated with high-density lipoproteins (HDLs),
albumin, and other plasma proteins in the blood; is incorporated by the
target cell through endocytosis; and accumulates mainly in the
lysosomes
23 and, in part, in the cell membrane and
membranous organelles (Obana et al., unpublished data, 2001).
There are also differences in the kinetics and tissue distribution
after in vivo administration. Whereas liposomal BPD-MA is rapidly taken
up by vascular endothelial cells and soon disappears from them, leaving
a large deposit of dyes in the retinal pigment epithelium
(RPE),
24 25 ATX-S10(Na) more gradually accumulates in the
neovascular wall with a peak at 1 hour, leaving little accumulation in
RPE.
19
In our recent study, for a clinical trial of PDT with ATX-S10(Na) in
humans, we determined the optimal timing and dose of irradiation for
the selective occlusion of experimental CNV in monkey
eyes.
21 The results demonstrated that laser irradiation at
30 to 74 minutes after injection of 4 or 8 mg/kg BW of ATX-S10(Na) and
at 30 to 150 minutes after injection of 12 mg/kg BW of dye effectively
closes CNV, with minimal damage to the healthy retinal and choroidal
vessels surrounding the lesion. Several previous studies using other
photosensitizers indicated that the optimal timing for laser
irradiation is early: 5 to 30 minutes after dye injection for
chloroaluminum phthalocyanine,
7 within 5 minutes for
mono-
l-aspartyl chlorin e6,
17 and 15 minutes
for BPD-MA.
15 In the preclinical studies of BPD-MA,
irradiation between 20 and 50 minutes has been suggested to be
appropriate.
10 The advantages of such early irradiation
might come from the lower doses of dye and irradiation required. It is
also reported that laser irradiation at later than 30 minutes causes
dye-leakage–related injuries in the sensory retina.
8 25 In this study, to determine whether PDT with ATX-S10(Na) has a similar
benefit in early laser irradiation, we compared the photodynamic effect
on CNV occlusion and the adverse effects such as injuries to retinal
vessels and recurrence of CNV between the PDT with early irradiation
(1–19 minutes after dye injection) in a low dose (2 mg/kg BW) of
ATX-S10(Na) and PDT with late irradiation (later than 30 minutes) in
high doses (4, 8, and 12 mg/kg BW).
21
Nine cynomolgus monkeys (2–2.5 kg) were used. They were treated
in accordance with the ARVO Statement for the Use of Animals in
Ophthalmic and Vision Research. All experimental procedures were
performed with monkeys under anesthesia with intramuscular injection of
50 to 60 mg/kg BW ketamine hydrochloride and 5 to 10 mg of diazepam.
Proparacaine HCl was used for topical anesthesia. Pupils were dilated
with 2.5% phenylephrine hydrochloride and 0.8% tropicamide.
Experimental CNV was induced in the posterior pole of the fundus
by photocoagulation with krypton laser (wavelength, 647 nm; Novus Omni
Laser; Coherent, Santa Clara, CA). At 14 to 31 days after
photocoagulation, CNV was confirmed by ophthalmoscopy, sodium
fluorescein (SF) angiography, and indocyanine green (ICG)
angiography, as previously described.
21
The photosensitizer 13,17-bis (1-carboxypropionyl)
carbamoylethyl-8-etheny-2-hydroxy-3-hydroxyiminoethylidene-2,7,12,18-tetraethyl
porphyrin sodium (ATX-S10(Na), Photochemical Inc., Okayama, Japan) is
an iminochlorine aspartic acid derivative. The dye was diluted with
distilled water into the concentration of 10 mg/ml before use. This dye
has an absorption peak at a wavelength of 401 nm in the Soret band and
at 664 nm in the Q band in a water solution. Both absorption peaks
shift to the longer wavelength in the plasma solution, and the peak in
the Q band in the plasma measures 670 nm. When the dye in the plasma
solution is excited with a 670-nm light, it emits fluorescence at a
maximum wavelength of 680 nm.
Six monkeys were divided into two groups. Each three animals
received an intravenous injection of 4 mg/kg and 8 mg/kg BW ATX-S10(Na)
as had been used in the previous study.
21 (Although the
dye dosage of 2 mg/kg was also tried, clear fluorescence images were
not obtained [data not shown].) The fluorescence of ATX-S10(Na) from
the fundus was detected by a modified fundus camera (TRC-50IA; Topcon,
Tokyo, Japan) which included a 670-nm wavelength diode laser (Hamamatsu
Photonics Inc., Hamamatsu, Japan) as the stimulation source and a
cooled charge-coupled device (CCD) camera (KP-160; Hitachi, Ibaragi,
Japan) as a detector. A sharp-cut filter (SC70; Fuji, Kanagawa, Japan)
blocked the light of wavelengths shorter than 670 nm and was used as a
barrier filter. Laser irradiance was 9.1 to 15.5
mW/cm
2, as measured on the corneal surface with a
power meter (Nova-Display; Ophir Optronics, Inc., Boston, MA). The
exposure time for taking angiographic photographs was less than 10
seconds, which was too short to cause photodynamic injury to the
retinal tissue. Autofluorescence was absent in the fundus before dye
injection. The data were recorded on S-VHS video tape and/or a digital
video system. Angiograms were performed on both eyes at 2, 4, 6, 8, 10,
12, 14, 16, 18, 20, 30, 40, 50, and 60 minutes and 1.5, 2, 2.5, 3, 3.5,
4, 6, 8, and 24 hours after dye injection.
PDT was performed on 45 lesions of CNV in six eyes in three
monkeys under general anesthesia. At 1 to 19 minutes after intravenous
injection of 2 mg/kg BW ATX-S10(Na), laser irradiation of 6.9 to 437.5
mW/cm2, as calculated on the retinal surface,
which does not induce thermal coagulation, was conducted for 30 to 240
seconds (radiant exposure was calculated to be 1–86.9
J/cm2), using a slit lamp system equipped with a
670-nm diode laser (Hamamatsu Photonics Inc.) and a fundus contact lens
(IF-210R; Menicon, Nagoya, Japan). The spot size on the retinal surface
was 1500 to 3000 μm in diameter, which covered the whole area of CNV.
Laser power was checked before every experiment by a power meter
(Fieldmaster; Coherent). As controls, two lesions of CNV from two
monkeys were subjected to dye injection without irradiation, and three
lesions from two monkeys were irradiated by a laser at 437.5
mW/cm2 (61.3 J/cm2) for 140
seconds before dye injection.
PDT was performed on five areas of the normal chorioretina in
two eyes of one monkey. At 1, 3, 5, 9, and 11 minutes after intravenous
injection of 2 mg/kg BW ATX-S10(Na), laser irradiation of 437.5
mW/cm2 was conducted for 60 to 120 seconds with a
radiant exposure of 26.3, 26.3, 32.8, 39.4, 52.5
J/cm2, respectively, with a spot diameter of 2000μ
m.
One day after PDT, vascular occlusion was identified by SF and
ICG angiography. Observations were made every week for healthy and
neovascularized regions until 3 and 4 weeks after testing,
respectively. Monkeys were then killed by intravenous injection of an
overdose of pentobarbital sodium, and the eyes were enucleated and
fixed in Karnovsky fixative overnight at 4°C. Semi-thin sections were
stained with toluidine blue and observed by light microscopy.
In ATX-S10(Na) angiography, bright fluorescence appeared in the
choroid at 6 seconds after dye injection. Dye then appeared in the
retinal arteries on the optic disc and in the venous return within the
next few seconds. The choroidal vessels were more clearly seen by
ATX-S10(Na) angiography
(Fig. 1A) than by SF angiography
(Fig. 1C) . Dye infusion in the short and long
posterior ciliary arteries, however, was less obviously demonstrated by
ATX-S10(Na) angiography
(Fig. 1A) than by ICG angiography
(Fig. 1D) .
Hyperfluorescence in ATX-S10(Na) angiography representative of CNV
lesions was observed within 10 seconds and gradually increased in
intensity
(Fig. 1A) , although the dye leakage shown by ATX-S10(Na)
angiography was less obvious with SF angiography
(Fig. 1C) . At 10 to 20
minutes, ATX-S10(Na) fluorescence in the retinal capillaries had faded
away, whereas that in the retinal arteries and veins persisted.
Hyperfluorescence in the CNV was still apparent, showing a mild degree
of dye leakage (data not shown). At later than 30 minutes, whereas dye
fluorescence in the retinal arteries and veins was diminished, that in
the CNV persisted until 90 minutes and 120 minutes, with dye doses of 4
and 8 mg/kg, respectively
(Fig. 1B) , and then declined at 180 minutes.
By 24 hours, ATX-S10(Na) fluorescence entirely diminished in all the
structures of chorioretinal tissues.
Effects of PDT.
By ophthalmoscopy, the CNV present before PDT was seen as yellowish
gray subretinal proliferative tissue (data not shown) with ring-shaped
hyperfluorescence showing late dye leakage in SF angiography
(Fig. 2A) and ring hyperfluorescence with negligible amounts of late dye leakage
in ICG angiography (data not shown). Immediately after PDT, no obvious
changes were found in ophthalmoscopy. At 1 day after irradiation, 22 of
45 irradiated CNV lesions were occluded as indicated by whitish opacity
with a grayish white halo in the full thickness of retina shown by
ophthalmoscopy
(Fig. 2B) and disappearance of ring hyperfluorescence
with late dye leakage shown by SF angiography
(Fig. 2C) and ICG
angiography
(Fig. 2D) . Choriocapillary occlusion, as indicated by
hypofluorescence in SF angiography
(Fig. 2C) and ICG angiography
(Fig. 2D) , was concurrently detected. Moreover, damage to the blood–retinal
barrier in the retinal pigment epithelial (RPE) cells occurred, as
represented by fluorescein leakage at the margin of lesions in SF
angiography
(Fig. 2C) . In contrast to choriocapillaries and CNV, large
choroidal vessels were well perfused in ICG angiography
(Fig. 2D) .
In 20 of 22 CNV-occluded lesions, the retinal arterioles and venules
were not occluded but exhibited fluorescein leakage from the vessels,
whereas the retinal capillaries were closed
(Fig. 2C) . In the remaining
two lesions, no damage to retinal vessels was seen. Among 23 CNV
lesions without CNV closure, some lesions with mild opacity in
ophthalmoscopy displayed fluorescein leakage as the result of a break
in the blood–retinal barrier in the RPE, whereas other lesions with no
opacity showed no apparent changes in SF and ICG angiography. At 1 week
after irradiation, retinal opacity decreased in intensity in the
lesions, and dye leakage disappeared from the retinal arterioles and
venules in SF angiography. However, at 1 to 4 weeks, ring-shaped
hyperfluorescence with late dye leakage, which represented recurring
neovascularization, appeared in some lesions in SF
(Fig. 3A) and ICG
(Fig. 3B) angiography. Histologic analysis demonstrated that
new vessels were formed in the subretinal proliferative tissue and that
underlying choriocapillaries were not occluded (
Fig. 4A , arrow in inset). In the irradiated region adjacent to the subretinal
proliferative tissue, pigment-laden cells overlay the RPE
(Fig. 4B) .
In control eyes subjected to laser irradiation alone or dye
administration alone, there were no appreciable ophthalmoscopic or
angiographic changes.
Figure 5 summarizes the efficacy of the PDT that was conducted at 1 to 20
minutes after administration of 2 mg/kg BW of ATX-S10(Na). Among 22
CNV-occluded lesions, 20 lesions displayed damage to retinal arterioles
and venules. It was therefore difficult to clearly delineate the zone
of optimal treatment conditions for selective CNV occlusion. Moreover,
CNV recurred in 19 lesions, including the two with selective CNV
occlusion.
To more clearly demonstrate the adverse effect of PDT on the
normal tissue surrounding neovascular lesions, we examined PDT-induced
changes in normal eyes. Immediately after PDT, irradiated lesions
showed no changes on ophthalmoscopy. At day 1, a color change from
grayish to whitish was observed in ophthalmoscopy. Treatment conditions
that induced CNV closure in the above experiments caused
hypofluorescence indicative of choriocapillary occlusion in the normal
eyes in early-phase SF and ICG angiography in four of five lesions
(Fig. 6) . Fluorescein leakage from the retinal arterioles and venules was also
noted. Hyperfluorescence was induced in all the irradiated lesions in
late-phase SF angiography (data not shown). One lesion that was treated
with a laser exposure of 26.3 J/cm
2 3 minutes
after dye injection showed the occlusion of large choroidal vessels in
ICG angiography (data not shown).
At 1 week, although fluorescein leakage from the retinal arterioles and
venules was no longer seen, choriocapillary occlusion persisted. At 2
to 3 weeks, PDT-treated lesions showed pigment mottling in
ophthalmoscopy. Choriocapillary occlusion, as represented by
hypofluorescence in early-phase SF angiography, was no longer seen, and
mottled hyperfluorescence and hypofluorescence were noted instead. In
late-phase ICG angiography, hypofluorescence was observed in treated
areas. Histology demonstrated that RPE preserved its original structure
of a single cell layer (
Fig. 7 , inset). Many macrophages were accumulated on the apical side of the
RPE, representing extensive inflammation. They vigorously incorporated
pigments, which may have interfered with the fluorescence of ICG. The
choroid showed normal architecture with patent choriocapillaries.
Retinal vessels were also open. Outer segments of photoreceptors became
shortened or were absent in part. Most of the nuclei in the outer
nuclear layer were weakly stained with toluidine blue, and some showed
pyknotic changes. Nerve fibers and ganglion cell layers were normal in
appearance.
In PDT for cancer, to obtain a selective effect, laser irradiation
is conducted at the time point when photosensitizers are more
preferentially accumulated in the cancerous tissue than in the
surrounding normal tissue. For instance, in PDT with hematoporphyrin
derivatives, the optimal time point for irradiation is 48 to 72 hours
after dye injection when tumor cells and endothelial cells of tumor
capillaries vigorously incorporate dyes. In the ophthalmology, however,
there remains controversy about whether laser irradiation immediately
or at some interval after dye injection exerts a selective injuring
effect on CNV. The proposal of immediate irradiation comes from the
belief that high concentrations of blood-borne dye achieved at an early
time has the maximal photodynamic effect on vascular endothelial cells
by generating a large amount of free radicals, even with low doses of
photosensitizer and irradiation.
18 In fact, in PDT with
chloroaluminum sulfonated phthalocyanine and
mono-
l-aspartyl chlorin e6, laser irradiation at 5 to 30
minutes and 5 minutes, after dye injection, respectively, yields the
maximal occluding effect on CNV.
7 17 In PDT with BPD-MA as
well, irradiation is conducted at 5 minutes after the end of 10-minute
dye infusion.
15 Moreover, late irradiation may exert an
adverse effect on the RPE and sensory retina by causing dye leakage
from CNV and the choriocapillaris.
7 8 16 24 26
We previously observed in PDT with ATX-S10(Na) that laser irradiation
at 30 to 74 minutes after injection of 4 or 8 mg/kg BW of dye and at 30
to 150 minutes after 12 mg/kg BW of dye induced a highly selective CNV
closure with minimal damage to surrounding healthy retinal and
choroidal vessels.
21 As demonstrated in this study, early
irradiation (1–19 minutes) with a low dose (2 mg/kg BW) of
ATX-S10(Na), however, failed to achieve selective CNV occlusion without
neovascular recurrence; 43 of 45 conditions did not induce selective
CNV closure (although concurrent damage to retinal vessels and normal
chorioretina was repaired within 2–3 weeks) and the remaining two
conditions induced selective CNV occlusion but caused CNV recurrence.
The selectivity of PDT was closely related to the biodistribution and
kinetics of dye accumulation in the tissue, which seem to depend on the
chemical properties of dye. In this study, ATX-S10(Na) angiography
demonstrated that, at the dose of 8 mg/kg BW, dye accumulation in CNV
persisted until 120 minutes, whereas that in the retinal capillaries
and retinal arteries and veins persisted until 30 minutes, indicating
that the dyes accumulated more in CNV between 30 and 120 minutes. This
time interval was compatible with that of laser irradiation for
selective CNV occlusion. At a dose of 2 mg/kg BW, although no clear
fluorescence images were obtained by angiography (data not shown),
distribution and kinetics of dye accumulation was considered to be
comparable. Nonselectivity of the injuring effect of early PDT with 2
mg/kg BW ATX-S10(Na) was considered to result from the absence of
preferential accumulation of dye in CNV lesions at this time point. It
was also noted that the dye leakage from the CNV was less appreciable
in ATX-S10(Na) angiography than in SF angiography, which may be related
to the higher affinity of ATX-S10(Na) to the plasma proteins. The
property of little leakage from the CNV will decrease the adverse
effect of PDT on the sensory retina.
The recurrence of CNV is a major clinical problem in
PDT.
14 15 This pathologic event is considered to involve
not only recanalization but also regrowth, because the recurring CNV
lesion is usually larger than the original lesion, as shown in
Figures 2A and 3A . In the present study, CNV recurrence occurred in 19 of 22
CNV-occluding conditions and was usually seen in the lesions where
retinal injuries (i.e., closure of capillaries, damage to the retinal
arterioles and venules, and a break of the blood–retinal barrier) were
extensive. Tissue injuries and the hypoxia in the sensory retina induce
angiogenesis through macrophage accumulation and vascular endothelial
growth factor (VEGF) production.
27 28 29 Proliferation of
VEGF-expressing fibroblasts has been demonstrated in
AMD.
30 In this study, early PDT provoked intense
inflammation in the retina, which may have led to a recurrence of the
CNV, probably mediated by VEGF.
Between the human AMD and monkey CNV examined in this study, there must
be differences in pathogenesis; the former disorder is degenerative,
whereas the latter includes potent inflammatory responses. The optimal
condition for selective PDT in human AMD may accordingly differ from
that obtained in our present and previous studies
21 and
therefore should be determined in a future clinical trial. The present
data, however, strongly suggest that, when applied in human AMD, early
laser irradiation (within 20 minutes after injection), even with
low-dose ATX-S10(Na), similarly fails to achieve selective CNV
occlusion and produces a higher incidence of CNV regrowth. Laser
irradiation at later than 20 minutes is thus recommended.
The authors thank Isao Sakata, Photochemical Inc., Okayama, Japan,
for providing the ATX-S10(Na).