Targeting of the Retinal Pigment Epithelium (RPE) by Means ...

Lasers in Surgery and Medicine 32:252–264 (2003)

Targeting of the Retinal Pigment Epithelium (RPE)by Means of a Rapidly Scanned ContinuousWave (CW) Laser Beam

Ralf Brinkmann,1* Norbert Koop,1 Mustafa Ozdemir,1 Clemens Alt,1 Georg Schule,1

Charles P. Lin,2 and Reginald Birngruber11Medical Laser Center Lubeck, 23562 Lubeck, Germany2Wellman Laboratories of Photomedicine, Massachusetts General Hospital, Harvard Medical School,Boston, Massachusetts 02114

Background and Objectives: Selective treatment of theretinal pigment epithelium (RPE) by repetitively applyinggreen ms-laser pulses is a new method for retinal diseasesassociated with a degradation of the RPE, which spares theneural retina. We investigated an alternative approach torealize repetitive ms-laser exposure by rapidly scanning acontinuous wave (CW)-laser beam across the RPE.Study Design/Materials and Methods: An Arþ laserbeam (514 nm) with a diameter of 18.75 mm was repetitivelyscanned across porcine RPE samples in vitro providing anirradiation time of 1.6 ms per point on the central scan axis.RPE cell damage was investigated by means of thefluorescence viability assay Calcein-AM.Results: The ED50 cell damage is 305 mJ/cm2 whenapplying 10 scans with a repetition rate of 500 Hz. Thethreshold decreases with the number of scans, a saturationwas found at 135 mJ/cm2 with more than 500 exposuresapplied. The depth of focus in beam direction is 350 mm,defined by an increase of the threshold radiant exposureby 20%.Conclusions: Targeting of pigmented cells with high localresolution has been proved with a laser-scanning device.Looking ahead selective RPE-treatment, the adaptation ofa laser-scanning device on a slit-lamp or into a modifiedretina angiograph seems to be an attractive alternative tothe pulsed ms laser device. Lasers Surg. Med. 32:252–264,2003. � 2003 Wiley-Liss, Inc.

Key words: fluorescence microscopy; melanosome; laserscanner; RPE damage; selective treatment; photocoagula-tion; viability assay

INTRODUCTION

The concept of selective targeting of naturally or arti-ficially pigmented cells or organelles in less strong absorb-ing surroundings was introduced by Anderson and Parrish[1,2] and led to a variety of applications in ophthalmology[3,4] and dermatology [2] using pulsed laser radiation.Selective cell effects are most interesting in cases wherehighly sensitive cells, which have to be preserved, are closeto the target cells. As an example of such an application,in vitro targeting of the retinal pigment epithelium (RPE)in the fundus of the eye has been investigated in this study

by means of a scanning application using continuous wavelaser radiation.

A variety of retinal diseases such as diabetic macularedema, drusen, and central serous retinopathy are thoughtto be caused by a decreased function of the RPE. A methodfor selective destruction of RPE cells without producingadverse effects to adjacent tissue, especially to the neuralretina and the choroid, seems to be an appropriate methodof treatment [5], presuming that the destroyed cells will bereplaced by proliferation or migration of neighboring intactRPE cells.

Selective RPE damage can be obtained due to the factthat the melanosomes inside the RPE cell absorb about 50%of the incident light in the green spectral range [6] andthus are the dominant chromophores within the fundusof the eye. When laser pulses with appropriate energy areapplied, high temperatures are induced at the melano-somes. In case of ms pulse duration, the high temperature isconfined to the RPE and leads to the selective RPE celldamage, while only a low sublethal temperature increase isobtained in the adjacent tissue structures [5]. With respectto the origin of cell damage, we refer to Discussion.

The selective damage of the RPE has first been demon-strated by Roider in rabbits by using 10–500 Argon-ionlaser pulses of 5 ms duration at a repetition rate of 500 Hz[4]. Fluorescein angiography was accomplished to visualizethe ophthalmoscopically invisible effects. Two weeks aftertreatment, the lesions were covered by a new population ofRPE cells. Four weeks post-treatment; a morphologicallycompletely restored RPE barrier showing normal RPE cellswas found [4]. The selectivity of damage to RPE cellssparing the photoreceptors was demonstrated by histologicexaminations at different times after treatment.

Contract grant sponsor: NIH (CPL); Contract grant number:EY12970.

*Correspondence to: Ralf Brinkmann, Medical Laser CenterLubeck GmbH, Peter-Monnik-Weg 4, D-23562 Luebeck, Germany.E-mail: [email protected]

R. Birngruber has disclosed a potential financial conflict ofinterest with this study.

Accepted 19 November 2002Published online in Wiley InterScience(www.interscience.wiley.com).DOI 10.1002/lsm.10150

� 2003 Wiley-Liss, Inc.

A first clinical trial using a train of ms-laser pulses from afrequency doubled Nd:YLF laser (wavelength 527 nm,pulse duration 1.7 ms, 30 or 100 pulses applied at arepetition rate of 100 or 500 Hz, respectively, on a retinalspot of 160 mm in diameter) has already proved the conceptof selective RPE effects and demonstrated the clinicalpotential of this technique [7,8]. Angiographical-ly determined threshold radiant exposures are around500 mJ/cm2 per pulse. For the treatment typically 650 mJ/cm2 are used, however, in order to compensate for thevariation in pigmentation and ocular transmission. Micro-perimetry showed no loss of retinal light sensitivity in thefollow-up period and demonstrated the functional selectiv-ity of this technique [9,10].

Since the selective effects in the RPE only rely on localhigh temperatures within these cells [11], this goal shouldalso be obtained with a continuous wave (CW) laser beam ofsufficient irradiance, which is rapidly scanned across theRPE. Combining a commonly available CW green laser, asit is used for routine retinal photocoagulation, with a sharpfocusing and fast scanning device, such a system might besuited to also induce selective RPE effects. An advantageof scanning laser over ms-pulsed application is the possibi-lity to induce RPE defects in an arbitrary pattern within theirradiated areas, thus investigating the therapeuticallymost useful irradiation geometry.

The aim of this study was to investigate whether scannedCW-laser irradiation can be used to induce RPE effects and

whether RPE-cell damage thresholds are similar to a largearea but pulsed irradiation, which are known from previouswork [11]. In that study, the damage thresholds of porcineRPE in vitro were determined for the following set ofirradiation parameters: 527 nm wavelength (frequencydoubled Nd:YLF-laser), 500 Hz pulse repetition rate, 50 mmlaser spot diameter with a top-hat beam profile. For pulsedurations of 1 and 3 ms and application of 10–500 pulses,ED50 threshold radiant exposures between 100 and160 mJ/cm2 were found [11] (compare Fig. 4) [11]. In thissetup, we used a small but also spatially top-hat beam pro-file with a scanning speed to provide a ms illumination timeon the RPE cells. Temperature calculations were performedfor a scanned spatially top-hat and Gaussian beam profilein order to assess possible clinical laser scanning devices.

MATERIALS AND METHODS

Experimental Setup

In order to evaluate RPE damage thresholds on porcineRPE samples, we used a setup as sketched in Figure 1. AnArþ-Laser (Spectra Physics, Inc., Mountain View, CA USA,model 2030-15s) operated at the wavelength of 514 nm witha maximum power of 7 W served as CW-laser source. Thebeam was coupled to a 25-mm core diameter fiber (FibertechGmbH, Berlin, Germany AS25, NA¼ 0.1). The length of thefiber was 200 m in order to reduce spatial intensitymodulations (speckle) at the distal fiber tip. The fiber tip

Fig. 1. Experimental setup of the laser scanning system including sample illumination. The

focal lengths f of the different lenses is given in millimeters. [Color figure can be viewed in the

online issue, which is available at www.interscience.wiley.com]

TARGETING OF THE RETINAL PIGMENT EPITHELIUM 253

was imaged onto the object plane with a magnification of0.75 according to the optical setup shown in Figure 1,yielding a theoretical spot diameter of 2o0¼ 18.75 mm. Inorder to minimize aberrations, a double telecentric 4-f-opticwas used in the scanned beam path. The scanning field forthis arrangement was 280 mm� 280 mm. The total power Pof the beam was measured with a power meter (Coherent,Inc., Santa Clara, CA USA).

An acousto-optic deflector (AOD) (Crystal Technologies,Palo Alto, CA USA AODF 4100-1, 50 MHz bandwidth)provides the fast line scan of 11.7 m/s on the sample toachieve an illumination time of 1.6 ms per point in thecentral line of the scan. The scanning speed was determinedby scanning the beam in the image plane across an objectscale and measuring the time between the fringes with aphotodiode mounted below. The complete line scan takes40 ms, while for only 24 ms the beam was on. The verticalscan was provided by a galvanometer scanner (GeneralScanning, Watertown, MA USA G120D) operated at a freq-uency of 500 Hz, which subsequently was the framing rateof the repetitive illumination. Since a complete field scanonly requires 640 ms, the nearly linear regime of thesinusoidal angle deflection curve can be used. The acousto-optic modulator (AOM) (Landwehr Electronic GmbH,Norderstedt, Germany, M85, 85 MHz) served to block theirradiation in unwanted areas such as the return path ofthe beam and between irradiated lines. The synchroniza-tion of the scanners and the number of illuminated lineswas provided by a control unit. The reproducibility toexactly irradiate the same lines with the following scans iswithin a lateral deviation of maximal 1 mm and in axialdirection of <0.5 mm.

Determination of the Radiant Exposureon the RPE

When scanning a circular beam of power P with a radiuso0 and a velocity v in direction z across the target, theradiant exposure H(x) at each point in the scan fielddecreases with distance x from the central scan axisaccording to:

HðxÞ ¼ 2P

p � o20 � v

�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðo2

0 � x2

qÞðtop hatÞ HðxÞ

¼ffiffiffi2

p

r� P

o0 � v� e

2x2

o2v ðGaussianÞ

ð1Þ

o0 represents the spot radius for a spatially top-hatprofile or the Gaussian beam radius, when applying aTEM00 laser beam. The irradiation time on the scan axis ist¼ 2o0/v. Thus, for example, to deliver a radiant exposureof H(0)¼ 100 mJ/cm2 at the scan axis (x¼ 0) with anillumination time of t¼ 1.6 ms and a scanning spot diameter2o0¼ 18.75 mm, a laser power of P¼ 173 mW for the top hatand P¼ 138 mW for the Gaussian beam profile is required.If a decrease of the maximum radiant exposure H(0) by 20%perpendicular to the scan axis is tolerated, a stripe with abroadness of 33% of the beam diameter is covered in theGaussian application mode, while it is 61% in the top-hatcase.

If laser light is transmitted by multimode optical wave-guides, the intensity at the distal fiber tip is spatiallymodulated, owing to coherent interference of different fibermodes propagating on different fiber path lengths. Here,the intensity distribution at the fiber tip is imaged onto theobject plane. Due to beam deflection and aberrations alongthe long light path in the scanner, a non-ideal image of thefiber tip is obtained at the probe. The spot diameter and itsintensity distribution at the object plane were measuredwith a CCD-camera using a 40� magnification, withoutoperating the scanner. The beam profile was analyzed witha beam analyzing software (Polytec GmbH, Waldbroon,Germany, LBA-100).

Figure 2a shows a typical intensity profile, which isscanned across the RPE. Due to spatial aberrations andthe intensity modulation, equation 1 cannot be used toproperly calculate the radiant exposure H(x) each point inthe scanned field experiences. An accurate way to deter-mine the radiant exposure for a scanned beam of arbitraryshape and modulation is described below and used in thisstudy: If a beam with an intensity profile I(x,z) is scannedwith the speed v in z-direction over the probe, any point inthe scan field experiences the radiant exposure Hscan(x)according to:

HscanðxÞ ¼1

v

ZIðx; zÞ � dz ð2Þ

Using the intensity distribution I(x,z) achieved from theCCD-image (standing beam), Hscan(x) can be achieved bynumerically summarizing the intensity in z-direction:

HscanðxÞ ¼1

v

Xz

I x; zð Þ � Dz ð3Þ

by normalizing the sum of all camera pixel values F(x,z) tothe total incident power P using

I x; zð Þ ¼ PPpixel

F x; zð Þ �F x; zð ÞDx � Dz

ð4Þ

Dx times Dz denote the sizes of a camera pixel, regardingthe optical magnification. Figure 2b shows the calculatedradiant exposure achieved by summarizing the pixel in-tensities, shown in Figure 2a, according to equations 3and 4. The maximal radiant exposure Hmax, which is notnecessarily on the central scan axis, is defined by

Hmax ¼ max HscanðxÞð Þ ð5Þ

The ratio f of this experimental determined value Hmax

to the maximal radiant exposure H(0) of the ideal top-hatbeamprofile according to equation 1 using 2o0¼18.75mm, is:

f ¼ Hmax

Hð0Þ Hmax ¼ f � 2 � P

p � o0 � vð6Þ

Porcine Eye Model and Viability Assay

RPE tissue samples were harvested from freshly enu-cleated porcine eyes. Stored in cold saline solution, porcineRPE cells survive several hours post mortem. RPE sampleswere prepared and used for the experiments within 4 hours

254 BRINKMANN ET AL.

Fig. 2. a: The image shows a 3-D intensity plot of the real beam

profile at the image plane (specimen). b: The solid trace shows

the radiant exposure, which each point in the scan field

experiences, depending on the distance x of the observation

point to the central scan axis. The radiant exposure is obtained

by integrating the intensity over time, assuming a scanning

speed v¼ 11.7 m/s of the spot in scan direction (z-direction).

Additionally plotted (dashed line) is the radiant exposure,

which a non-modulated ideal top-hat beam with a diameter of

18.75 mm, would provide. It gives a parabolic profile according

to equation 1.


post mortem. One circular section of about 1 cm2 from thefundus of each eye was taken, including retina, sclera, andchoroid. The number of sections used to evaluate thethreshold data points is indicated in Figure 4. Prior toirradiation, the neural retina was carefully peeled off.

RPE cell viability was probed by the dye Calcein-AM(Molecular Probes, Inc., Eugene, OR USA) that tags livingcells. In living cells, Calcein-AM is reduced by esterases toCalcein, which is the actual fluorescent marker. Calceincan be excited with blue light (excitation maximum at 490nm) to fluoresce in the green spectral region. Calcein-AMwas used in a phosphate buffer solution at a concentrationof 2 mg/ml. Twenty microliters were dropped onto the RPEcell layer directly after irradiation. Test exposures prior tothe study showed the same ED50 cell damage thresholds,independent whether Calcein-AM was applied to thesample prior or past irradiation. However, the fluorescencecontrast between death and damaged cells was higherwhen staining after irradiation, since it prevents fluores-cence bleaching by the Argon-laser beam. The sample wascovered with a microscope cover slip in order to preventdesiccation. The probe was observed 30 minutes afterirradiation with a fluorescence microscope. Living cellsfluoresce, damaged cells appear dark as exemplary shownin Figure 3b.

Irradiation Modalities and DamageThreshold Determination

The sample was illuminated from the scleral side withwhite light. The image plane of the scanner and the CCDcamera were adjusted to each other. A sharp image of theRPE on the CCD chip was used to adjust the sample to theimage plane of the scanner. After sample adjustment,carefully choosing flat areas, four scanning lines wereperformed on each area as shown in Figure 3b. For most ofthe RPE probes, several scan fields could be applied ontoone sample. To easily determine damage threshold irra-diances of the cells, a power gradient along each scannedline, as shown in Figure 3a, was provided by linearlydecreasing the diffraction efficiency of the acousto-opticdeflector over the length of a scan line.

The cell damage thresholds were determined by measur-ing the length of the line of damaged cells, dth, with respectto the length of the complete scanned field, dfield¼ 280 mm.For evaluation, at least, two damaged cells have to be incontact, while single isolated dead cells, which were oftenfound randomly distributed on the sample, were notconsidered. The threshold power Pth was determined bycorrelating the line of damaged cells with the powergradient according to Figure 3, using a linear approxima-tion for the power gradient:

Pth ¼ Phigh � dth

dfield� Phigh � Plow

� �ð7Þ

The ED50 damage radiant exposure Hth was determinedby inserting Pth in equation 6 yielding

Hth ¼ 2f

p� Pth

o0 � v¼ f

p� Pth � t

o20

ð8Þ

For each parameter set, samples from 10 different eyeswere used. Per eye, several RPE samples could be prepared,which were irradiated at 3–6 different power levels. For theED50 cell damage radiant exposure referred to in Figures 4and 5, the mean values with standard deviation werecalculated. In order to measure the threshold dependencewith respect to deviations of the probe from the image planeof the scanner (proper focusing), the RPE was cut in stripesto achieve large areas of most flat RPE.

Temperature Calculations

Temperature distributions at the melanosomes werecalculated using an analytical solution of the equation forheat diffusion in a spherical particle [12]. This model wasdescribed by Thompson [13] and in detail with respect to msirradiation in a previous paper [11]. Melanosomes aremodeled by spheres with a diameter of 1 mm having thethermal properties of the surrounding water. It is assumedthat inside the sphere heat is produced spatially homo-geneously throughout the volume at constant rate duringthe laser pulse. The total energy, which is absorbed by amelanosome during irradiation, was calculated by assum-ing an absorption coefficient of 0.8 mm�1 [11]. Correctionsfor the higher index of refraction of melanin, Mie scattering,and coherent effects due to the small size of the particleswere not made. Light absorption in the choroid wasneglected. All calculations were performed on a PC usingMathematica 3.0.

RESULTS

A typical plot demonstrating the intensity distributionI(x,z) in the beam profile at the image plane is shown inFigure 2a. Figure 2b shows the radiant exposure, whicheach point in the scan field experiences, related to itsdistance x to the central scan axis, using equations 3–5.The parabolic profile is a plot of the ideal top-hat profileaccording to equation 1. The ratio f¼ 0.90 � 0.06, asdefined in equation 6, was determined as an average valuewith standard deviation evaluated out of 20 CCD frames.It is interesting to note that f< 1, even though speckleformation takes place, which normally leads to f > 1. Thereason for the small f is spherical aberration resulting fromthe non-ideal image of the distal fiber tip, which leads tomuch a broader beam as the theoretical beam diameter of2o0¼ 18.75 mm in the image plan, as demonstrated inFigure 2.

Figure 3b shows an example of the scan field influorescence microscopy with four illuminated scan lines,the power gradient as shown Figure 3a was applied alongthe scan direction. The irradiation was performed with100 scans. The mesh of RPE cells is clearly visible; thedark, non-fluorescing cells were defined as damaged.The length of the damaged line of cells was evaluated todetermine the threshold power according to Figure 3a andequation 7.

Figure 4 shows the cell damage ED50 radiant exposurefor applying 10, 50, 100, 300, 500, 750, and 1,000 repeti-tive exposures to the RPE. Applying 10 exposures, Pth¼569 mW was found, which results in an ED50 threshold


radiant exposure of Hth¼ 297 mJ/cm2, as calculated usingequation 8. The threshold strongly decreases to Hth¼131 mJ/cm2 when applying 500 scans, while a highernumber of scans does not lead to a further thresholddecrease. The solid line in Figure 4 represents a least

square fit of the threshold data according to the empiricalfunction ED50(N)¼ED50(1)�N�0.25, which is commonlyused to determine the additive effect of multiple exposures.N refers to the number of exposures applied, ED50(1)¼584 mJ/cm2 is calculated. For comparison, the threshold

Fig. 3. a: A power gradient was applied in scan direction. The

complete scan time over the length of a line, dfield¼ 280 mm,

takes 24 microseconds, the peak power (in this case 535 mW) is

reduced by 43% over the scan line. For a line of damaged cells

with the length dth, the threshold power Pth was determined as

sketched.b: Typical scan of damaged RPE cells by applying 100

exposures (complete dark hexagons) using fluorescence micro-

scopy. The scan direction was from the left to the right side.


Fig. 4. ED50 RPE threshold radiant exposures as a function of number of exposures for the

scanned mode (upper curve) and the pulsed mode with pulse durations of 1 and 3 ms according

to [11], respectively. A N�1/4-least square fit was plotted through the scanner data. The

numbers indicated in the figure refer to the number of samples (¼number of eyes) used to

evaluate the data point. [Color figure can be viewed in the online issue, which is available at

www.interscience.wiley.com]

Fig. 5. Laser power at RPE damage threshold depending on the distance of the probe to

the image plane.


data achieved with the pulsed setup using 1 and 3 ms pulsedurations are also shown [11].

The distance between the sample and the image plane ofthe scanner was varied in order to determine the damagethreshold laser power with respect to proper focusing in theaxial direction. This dependence is shown in Figure 5 for100 exposures per irradiated area. It shows a nearlyconstant threshold laser power of 420 mW over a range ofDz¼ 250 mm, corresponding to Hth¼ 218 mJ/cm2 on thescan axis. Further deviations result in a strongly increasingthreshold, 50% in a zone only 100–150 mm outside thecentral 250 mm zone. In order to enable proper adjustment,very lightly pigmented RPE samples were used in thisexperiment, which explains the slightly higher thresholdradiant exposure compared to Hth¼ 173 mJ/cm2 achievedwith stronger pigmented RPE (Fig. 4).

Figure 6 is a plot of the temperature increase at the rim ofa melanosome on the scan axis for a beam diameter of 18 mmand a scanning speed of 11.25 m/second. The temperatureincreases towards its maximum at the end of the irradia-tion in case of a spatial top-hat beam profile. For a Gaussianprofile of the same radius (1/e2) and laser power, a 35%higher temperature is calculated, the maximal tempera-ture is achieved about 400 nanoseconds after the beamcenter has passed the observation point. With respect to theabsolute temperature, it shows that a power of 500 mWunder top-hat irradiation conditions is sufficient to induce atemperature increase DT > 1008C at the melanosomes.

DISCUSSION

The ED50 threshold radiant exposure for RPE celldamage by use of a laser scanner has been investigated.A circular spot was repetitively scanned across a porcineRPE layer invitro withanequivalentexposure timeof 1.6msper point on the fast scan axis with a framing rate of 500 Hz.The threshold irradiances strongly decrease with increas-ing number of exposures. However, it is most remarkablethat the damage thresholds are significantly higher than inthe pulsed application mode under similar irradiationconditions [11]. These results, in comparison to thoseachieved in the pulsed mode, are analyzed and discussedin the following paragraphs. Furthermore, we describe anddiscuss possible clinical scanner based treatment systemsand judge their features with respect to the pulsed ms laserapplication system.

Origin of Cell Damage

When laser pulses with appropriate energy are applied toRPE cells, high temperatures are induced in and aroundthe melanosomes. If the vaporization temperature of theintracellular plasma is reached, which takes place at first atthe surface of the melanosomes, microbubbles begin to formhere. Due to the simultaneous growth of a high numberof microbubbles, the cell volume transiently increases,subsequently disrupting the cell structure. Recent investi-gations demonstrated this effect with nanosecond pulses:Lin et al. experimentally observed microbubble induced

Fig. 6. Calculations of the temperature increase at the rim of

an isolated melanosome in suspension on the central scan axis

(x¼ 0) for scanned Gaussian and top-hat beams with a

diameter of 18 mm, a laser power of 500 mW, and a scanning

speed of 11.25 m/s, respectively. The dashed line marks the

temperature increase when the center of the beam passes the

observation point. The temperature increase is directly

proportional to laser power applied. [Color figure can be viewed

in the online issue, which is available at www.interscience.

wiley.com]


selective cell killing in pigmented cells with nanosecond-pulses [14]. Kelly showed bubble formation inside bovineRPE cells and found that the threshold radiant exposuresfor bubble formation and for cell death are similar for singlepicosecond-laser pulses [15]. A recent study with varyingpulse durations in the nanoseconds–ms time range sup-ports the thesis of microvaporization induced RPE celldamage [11].

In order to prove, whether the vaporization thresholdmight be reached in this approach, temperature estima-tions were performed for the surface of a single melanosomein water suspension. Figure 6 shows the course of thetemperature with a peak increase of DT¼ 1088C, whenusing a 500 mW beam with a diameter of 2o0¼ 18 mm andan exposure time of 1.6 ms. With respect to the cell damageevaluated here: Hth¼ 297 mJ/cm2 (10 exposures) corre-sponds to an equivalent laser power of 472 mW according toequation 1, thus gives a peak temperature of T¼ 1248C,incl. room temperature of 228C. Hth¼ 130 mJ/cm2 (>500exposures) leads to T¼ 678C.

However, for more accurate temperature calculations,the tight mesh of melanosomes inside the RPE cell has tobe taken into account. It has been shown that the temp-erature at a melanosome inside a dense mesh strongly addsup during irradiation due to heat flow from adjacentmelanosomes, if pulse durations exceeding the thermalrelaxation time of a single melanosome, tR� 400 ns, areused [11].

In conclusion, temperatures above the vaporizationthreshold of water are most likely induced at cell damagethreshold at the melanosomes with the laser parametersused here. Therefore, we assume thermo-mechanical celldisruption owing to microbubble formation as the origin ofRPE cell damage and use this thesis in the followingsections of Discussion. An overview about other pos-sible mechanisms of RPE cell damage in the ms-timedomain is given in the literature but seem to be moreunlikely [5,11].

Threshold Dependence on the Numberof Exposures

The experiments showed that the threshold radiantexposure for cell damage significantly decreases with thenumber of exposures, possible reasons shall be discussedhere.

For a variety of pulse durations, pulse numbers, andwavelengths it has been shown in the literature that thedamage threshold scales with N�1/4, when N expositionsare applied [16]. This empirically found dependence hasbeen verified by a number of studies on retinal laser damagethresholds [17,18]. Applying this law to our experimentaldata, the solid line shown in Figure 4 represents a quitegood accordance to this law between 10 and 500 scans. Withrespect to the damage threshold for only one scan, whichcould not be investigated due to too less laser poweravailable, a radiant exposure of 584 mJ/cm2 can be cal-culated. This is nearly two times higher than that for10 scans, which seems to be very unlikely, if we compare itto the pulsed data shown in Figure 4 with about 30%

threshold increase. Further, the radiant exposure satura-tion as observed for higher number of exposures does alsonot obey this law. In conclusion, the N�1/4 law can only beapplied over a very limited range of 10–500 exposures.

The threshold saturation can also not be explained withthe Arrhenius formalism for thermal damage, since it alsopredicts a steady decrease of the thresholds for increasingexposures. It can also be excluded that an increase of theresidual RPE temperature prior to the next exposure leadsto a reduced threshold with increasing exposures: RPEtemperature estimations for a high number of exposureswith a spot diameter of 18 mm and a laser power of500 mW do not result in a significant residual temperatureincrease: DT< 58C after 500 pulses was calculated, thusthis effect can be neglected.

A more reasonable explanation for the lower thresholdradiant exposure with increasing number of exposuresmight be the statistical behavior of bubble formation atthreshold. Lin proved that RPE cell damage can occur, ifonly one pulse out of a train of 100 pulses, 6 ms in duration,leads to vaporization.

Cell Selectivity

The selectivity of the method with respect to the neuralretina and the photoreceptors could not be investigatedwith this model and setup. However, Figure 3b shows thelateral selectivity of the irradiation: Even above threshold(begin of the scan line), neighboring RPE cells (typically10–15 mm in diameter), which were not passed by the laserspot, were not damaged. Thus, mechanical or thermaleffects owing to microbubbles or heat diffusion, respec-tively, originating from the irradiated cells are not suf-ficient to cause damage to neighboring cells. This selectivitysparing adjacent tissue is much likely also in the axial beamdirection towards the adjacent photoreceptors. For thepulsed irradiation, microperimetry performed on patientsshowed no loss of retinal light sensitivity in the follow-upperiod and demonstrated the functional selectivity of thistechnique [9,10], even though the radiant exposure wasmuch higher. It has also been proved by histologicexaminations at different times after irradiation of rabbits[4]. In conclusion, photoreceptor damage is most unlikelywith such a scanning irradiation mode, if the laser power iskept close above RPE damage threshold.

Threshold Radiant Exposures: Pulsed VersusScanning Mode

It is most remarkable that although the average radiantexposure for all points on the scan axis is the same in thepulsed and the scanning mode, the thresholds for thescanning irradiation are always higher, up to a factor of2 for 10 and 50 exposures. In order to understand thisdiscrepancy, a closer analysis of the different irradiationmodalities with respect to the origin of cell damageis needed:

1. In contrast to the pulsed mode, not all melanosomeswithin a cell experience the same radiant exposure, dueto the circular spot scanned, as shown in Figure 2. A


radiant exposure within a range of 20% of themaximum radiant exposure is achieved only in ascanned stripe of about 60% of the beam diameter,which is less than a typical cell diameter in this case.Further, most of the cells are not hit totally but onlypartially due to the fact that spot and the cell size areabout equal. This is supported by the observation thatat threshold, some damaged cells are often totallysurrounded by non-damaged cells.

2. Not all cell melanosomes experience the same radiantexposure at the same time. Most likely, microbubbleformation at threshold takes place at the end of theirradiation, when the highest temperature is justreached (Fig. 6). Thus, a front of a few small bubblesmoves through the cells at the very rear end of thescanned spot. Therefore, although the radiant exposureper melanosome is similar, the volume increase withinthe cell is much smaller in the scanning mode incomparison to the pulsed mode.

3. Since the beam profile is not an ideal top-hat profile, thehighest temperature is achieved before the beam hastotally passed the observation point, comparable to theGaussian profile (Fig. 5). Thus, the effective radiantexposure until the peak temperature is reached is cor-respondingly smaller as the values calculated forFigure 4, where the total applied energy was takeninto account.

All points most likely contribute to a higher thresholdradiant exposure in the scanned mode compared to thepulsed mode, thus the higher thresholds observed seemsreasonable. The intensity modulation within the spot dueto coherent interference might lead to an increase ordecrease of the threshold irradiance, depending on thelocal distribution of the speckle. The different spot diameterin both cases (50 mm pulsed and 18.75 mm scanned) shouldplay no role, since the temperature increase depends on theradiant exposure. An increase of the average RPE tem-perature due to repetitive exposures up to 500 Hz can beneglected in both cases.

Laser Scanner for Clinical Applications

The question arises how such a scanner system can berealized for a clinical application. In order to discuss scan-ning treatment modalities, let us first review the clinicallyavailable data for treatment with a pulsed laser by use of apulse duration of 1.7 ms (see Introduction): It shows that650 mJ/cm2 are commonly used for treatment, which isabout 30% above the angiographic threshold of about500 mJ/cm2. This is about four times higher than theporcine RPE damage threshold achieved with the samelaser using a viability stain as depicted in Figure 4, for 100pulses applied. Apart from the different methods to deter-mine RPE damage, the higher human threshold is mostlikely originating from two different facts: First of all, due tothe lower melanin pigmentation, the human RPE absorbs50–60% of light in the green spectral range, while porcineRPE absorbs nearly 100%. Secondly, the direct transmit-tance of light through the whole human eye has to be kept in

mind in comparison to the in vitro experiments on RPEspecimen. Although the total transmittance of light in thegreen spectral range is around 80%, the direct transmit-tance (within an angle of 18) is only 40% due to mostlyforward scattering of the ocular media [19].

Further referring to Figure 4, it shows that a 1.5 timeshigher radiant exposure is needed in the scanning modecompared to the pulsed mode in order to obtain RPEdamage, when applying 100 exposures, for reasons dis-cussed above. Assuming that this factor achieved withporcine eyes can be transferred to a clinical treatment, thana radiant exposure of about 750 mJ/cm2 is needed to reachthe angiographic ED50 threshold in humans by use of alaser scanner with an illumination time of 1.7 ms. About1,000 mJ/cm2 should be used for an efficient treatment.

With respect to the selectivity and thus safety of thetreatment, the so-called therapeutic window betweenangiographic and ophthalmoscopic visibility was investi-gated in rabbits, showing a safe treatment range between170 and 430 mJ/cm2 (527 nm, pulse duration: 1.7 ms, 100pulses at 500 Hz) [20]. Below this fluence, RPE damage isuncertain (angiographic ED50: Hth¼ 138 mJ/cm2), whileabove, extended damage becomes obvious (ophthalmo-scopic ED50: Hth¼ 480 mJ/cm2). Transferring this resultsto the clinical application, than �2,500 mJ/cm2 should notbe exceeded in order to avoid retinal damage.

In order to achieve a radiant exposure of 1,000 mJ/cm2

with a top-hat spot diameter of 18 mm and an illuminationtime of 1.6 ms along the scan axis, a laser power as high as1.6 W according to equation 1 is needed on the RPE, withoutregarding intensity modulation. However, reviewing equa-tion 8, there are several possibilities to reduce this highlaser power:

1. The spot diameter can be reduced, since the laser powerneeded to achieve the desired radiant exposure isinversely proportional to the spot diameter, if thescanning speed is kept constant (eq. 8). It drops eveninversely proportional to the square of the beamdiameter, if the illumination time per point is keptconstant. However, with respect to the limited opticalquality of the eye, the minimal achievable beamdiameter on the retina is about 10 mm. Further, withreduced spot diameter, the depth of focus is reduced,the consequences on this are discussed in more detailbelow.

2. A Gaussian instead of a top-hat beam profile can beused. In this case, about 25% less power is required toobtain the same temperature at the melanosomes onthe scan axis according to Figure 6. However, a smallerstripe of melanosomes experiences high temperatures(compare eq. 1), which might lead to an increaseddamage threshold, since less microbubbles are produced.

3. The illumination time can be increased, since itis reciprocally proportional to the laser power. How-ever, in this case, an increase of the damage threshold isexpected, as shown in Figure 4, since the temperatureincreases slower than proportional with time (Fig. 6)due to increasing thermal diffusion during irradiation


[11]. Further, the therapeutical window becomes sig-nificantly smaller. The safety range is reduced to a factorof 1.5 with an illumination time of 5 ms [21].

4. The number of exposures can be increased. In thiscase, it has to be assured that the same scan line can beirradiated several times, if an irradiation modus ofseparated scan lines is used. Therefore, micromove-ments of the eye have to be excluded, thus fundusmovements relative to the laser beam path have to bekept smaller than about 10 mm per 300 ms.

Slit Lamp Application

From the technical point of view, all conventional retinalphotocoagulations as well as the pulsed selective RPEtreatment are performed via a slit lamp with a multimodefiber adapted laser. The laser light is coupled to the eye witha contact lens, which is placed on the cornea to compensatefor the refractive power of the eye. In order to achieve ahomogeneous energy distribution on the retina, the distalfiber tip with its nearly top-hat intensity profile is imagedwith the slit-lamp optics onto the fundus. The axialadjustment Dz (z-adjustment) is performed manually bythe ophthalmologist by moving the slit lamp until thelaser spot of the pilot laser shows a sharp image on theretina. Under optimal conditions with a narcotized rabbit,we measured a reproducibility of focusing of aboutDz¼�200 mm, however, in practice, a much larger rangeis expected. Even if the slit lamp is properly used with anocular crossed hair to avoid accommodation mismatchbetween the ophthalmologist’s view and the image plane ofthe scanned spot, slight eye and patient movements canhardly be avoided. A further drawback is patients withbeginning cataract. Due to light scattering, proper focusingto the retina will become difficult and moreover, the scan-ning spot on the RPE is smeared out, thus damage thres-holds are expected to rise strongly. Therefore, cataractpatients have to be excluded from treatment. The problemof proper depth adjustment with respect to a safe buteffective treatment will be addressed in the followingparagraph.

For applying a top-hat beam profile, the analysis is quitecomplicated, since the beam profile changes in the near fieldaround the image. For an 18.75 mm spot diameter as usedhere, it is as small as 300 mm when allowing a maximalthreshold increase of 20% (Fig. 5). In order to simplify thediscussion, we focus on a Gaussian beam application, whichcan easily be calculated. The TEM00 focal beam radius o0

and the beam radius oz at a distance Dz to the focal planeare related by:

Dz ¼ p � o20

l�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffioz

o0

� �2s

� 1 ð9Þ

In order to analyze the change of the radiant exposureby defocusing, we assume that a radiant exposure of750 mJ/cm2 is required in order to guarantee treatment(25% less than in the top hat case as discussed above) byuse of a scanning speed v¼ 2o0/twith t¼ 1.7 ms. Combiningequations 8 and 9, the useful range of spot diameter with

respect to laser power and accuracy in depth of focus can becalculated.

Figure 7 shows the decrease in radiant exposure whendefocusing for different spot radii. The power was normal-ized to a radiant exposure of 1,300 mJ/cm2 in the focus,which is in the center of the therapeutic window. It showsthat with o0¼ 5 mm spot radius, a power of about 1Wis sufficient for RPE damage. But in order to stay withinthe therapeutic window, the focusing range is only�200 mm. If this range shall be increased, a larger spotdiameter is required. Using o0¼ 7.5 mm, the range isextended to � 500 mm, which can realistically be obtainedduring treatment; however, the power needed in this caseis above 2 W. The power further increases with spotdiameter. In conclusion, a beam diameter of 2o� 15 mmseems to be ideal. However, with an additional fast auto-matic z-tracker, a smaller spot diameter and less power canprobably be used.

Retina Scanner Application

Another possible application device makes use of theprinciple of a retina laser scanner (e.g., the HRA fromHeidelberg Engineering GmbH, Germany or the SLO fromRodenstock GmbH, Germany) commonly used for retinaldiagnostics. In this case, a collinear beam is illuminated onthe cornea, which focuses the laser beam to a minimal spotdiameter of approximately 10 mm onto the retina. In thiscase, axial deadjustment is uncritical and the requiredlaser power may be kept below 1 W for an exposure time of1.6 ms. However, lateral eye movements have still to beconsidered in order to make use of the threshold decreasewith multiple scans: Movements larger than the spotdiameter between the scans will lead to a strong increaseof the threshold (Fig. 4), since always different RPE istargeted. In the worst case, RPE-damage has to be providedwith a single scan. As a matter of principal, only slightmodifications (substantially a small laser and an optoa-coustic modulator to switch on the laser at the desiredtreatment locations) are needed to modify the retinascanner, since their intrinsic scanning rates are in theorder of the illumination times needed here. However, up tonow retina scanning systems have not been used fortreatment. The ophthalmologists have therefore to judgewhether a treatment in front of a monitor would beacceptable.

On-line Treatment Control System

A further point to be considered bases on the fact that theselective RPE defects are ophthalmoscopically invisible.Thus an automatic detection system reporting a successfulirradiation would be highly useful. The dosimetry iscurrently performed by fluorescein angiography posttreatment, which demarks the RPE barrier defect. By useof a combined RPE treatment and diagnostic system, such adetection could be realized either by autofluorescencechanges or by use of exogeneous chromophores appliedduring or directly after treatment. Moreover, other noninvasive detection methods such as optical [21] or optoa-coustical [22] on-line dosimetry control devices are under


investigation for the pulsed application, but have to befurther explored for a scanning device.

CONCLUSION

Damaging of pigmented cells under high local resolutioncould exemplary be demonstrated on the RPE with ascanning CW laser device. A spot diameter of 18.75 mm wasscanned with a speed of 11.7 m/s across the RPE using laserpowers around 500 mW. The ED50 thresholds for porcineRPE damage strongly decrease with the number of ex-posures applied. From 10 to 500 illuminations, theempirical N�0.25 law is an appropriate approximation,however, large deviations are found for higher numbers ofexposures. The thresholds are up to a factor of 2 above thecorresponding thresholds found for repetitively pulsedirradiation with the same illumination time.

In order to realize a clinical system for selective RPE-treatment, a modified diagnostic laser scanning devicewould be the first choice, since depth focusing is notrequired and thus a CW-laser power below 1 W can beexpected for an illumination time of about 1.7 ms. The powercan further be reduced by using a slower scanning speed,however, the therapeutic window with respect to RPE cellselective damage will become smaller. A slit lamp adaptedscanning unit is also feasible but probably requires anautomatic focusing control, if high laser powers shall beavoided. For both devices, lateral eye movements duringirradiation have to be kept smaller than the spot diameter,if separated RPE cells or lines of cells shall be targeted.

Otherwise, a strong increase of the RPE damage radiantexposure can be expected. The main advantages of ascanning laser system compared to a pulsed treatmentdevice is its high flexibility to induce arbitrary RPEtreatment pattern with a high local resolution. Thisapplication system would be optimally suited for investi-gating the most useful irradiation pattern for treatment ofthe different RPE related diseases.

REFERENCES

1. Anderson RR, Parrish JA. Selective photothermolysis:Precise microsurgery by selective absorption of pulsed laserradiation. Science 1983;220:524–527.

2. Parrish JA, Anderson RR, Harrist T, Paul B, Murphy GF.Selective thermal effects with pulsed irradiation from lasers:From organ to organelle. J Invest Dermatol 1983;80(6):75–80.

3. Latina MA, Park C. Selective targeting of trabecular mesh-work cells: In vitro studies of pulsed and CW laser interac-tion. Exp Eye Res 1995;60:359–372.

4. Roider J, Michaud NA, Flotte TJ, Birngruber R. Response ofthe retinal pigment epithelium to selective photocoagulation.Arch Ophthalmol 1992;110:1786–1792.

5. Roider J, Hillenkamp F, Flotte TJ, Birngruber R. Micro-photocoagulation: Selective effects of repetitive short laserpulses. Proc Natl Acad Sci USA 1993;90:8463–8647.

6. Gabel VP, Birngruber R, Hillenkamp F. Visible and nearinfrared light absorption in pigment epithelium and choroid.Congress Series: XXIII Concilium Ophthalmologicum 1978;450:658–662.

7. Roider J, Wirbelauer C, Brinkmann R, Laqua H, BirngruberR. Control and detection of subthreshold effects in the firstclinical trial of macular diseases. Invest Ophthalmol Vis Sci1998;39(4):104.

Fig. 7. Radiant exposure on the RPE versus axial defocusing

up to 1 mm using a focal illumination time of 1.6 ms

(defocussing¼ 0) on the central scan axis for different focal

spot radii applied by a slit lamp based scanner using a

Gaussian beam. The laser powers indicated on each curve are

normalized to a radiant exposure of 1,300 mJ/cm2 in the focus,

which is around the center of the indicated therapeutical

window for an illumination time of 1.6 microseconds. [Color

figure can be viewed in the online issue, which is available at

www.interscience.wiley.com]


8. Roider J, Brinkmann R, Wirbelauer C, Laqua H, BirngruberR. Subthreshold (retinal pigment epithelium) photocoagula-tion in macular diseases: A pilot study. Br J Ophthalmol2000;84:40–47.

9. Roider J, Wirbelauer C, Brinkmann R, Laqua H, BirngruberR. Variability of RPE reaction in two cases after selectiveRPE laser effects in prophylactic treatment of drusen.Graefe’s Arch Clin Exp Ophthalmol 1999;237:45–50.

10. Roider J, Brinkmann R, Wirbelauer C, Laqua H, BirngruberR. Retinal sparing by selective retinal pigment epithelialphotocoagulation. Arch Ophthalmol 1999;117:1028–1034.

11. Brinkmann R, Huttmann G, Rogener J, Roider J, BirngruberR, Lin CP. Origin of retinal pigment epithelium celldamage by pulsed laser irradiance in the nanosecond tomicrosecond time regimen. Lasers Surg Med 2000;27(5):451–464.

12. Carslaw HS, Jaeger JC. Conduction of heat in solids, secondedition. Oxford: Clarendon Press; 1959.

13. Thompson CR, Gerstman BS, Jacques SL, RogersME. Melanin granule model for laser-induced thermaldamage in the retina. Bull Math Biol 1996;58(3):513–553.

14. Lin CP, Kelly MW, Sibayan SAB, Latina MA, Anderson RR.Selective cell killing by microparticle absorption of pulsedlaser radiation. IEEE J select Topics Quantum Electron1999;5(4):963–968.

15. Kelly WM, Lin CP. Microcavitation and cell injury in RPEcells following short-pulsed laser irradiation. Proc SPIE1997;2975:174–179.

16. ANSI. 1993. Standard Z136.1: American Standard for theSafe Use of Lasers: American Standards Institute.

17. Ham WT, Mueller HA, Wolbarsht ML, Sliney DH. Evaluationof retinal exposures from repetitively pulsed and scanninglasers. Health Physics 1988;54(3):337–344.

18. Stuck BE, Lund DJ, Beatrice ES. 1978. Repetitive pulse laserdata and permissible exposure limits. Vol. Report No. 58.San Fancisco: Lettermann Army Institute of Research,Presidio of San Francisco.

19. Boettner EA, Wolter JR. Transmission of the ocular media.Invest Ophthalmol Vis Sci 1962;1:776–783.

20. Framme C, Schule G, Roider J, Kracht D, Birngruber R,Brinkmann R. Threshold determinations for selective RPEdamage with repetitive microsecond laser systems in rabbits.Ophthalmic Surg Lasers 2002;33(5):400–409.

21. Neumann J, Brinkmann R. Interferometric detection oflaser-induced microbubbles in the retinal pigment epithe-lium. In: Laser–tissue interactions, therapeutic applica-tions, and photodynamic therapy. van den Bergh, Hubert:Birngruber, Reginald; 2001;. Proc SPIE 4433:81–86.

22. Schule G, Huttmann G, Roider J, Wirbelauer C, BirngruberR, Brinkmann R. Optoacoustic measurements during ms-irradiation of the retinal pigment epithelium. Proc SPIE2000;3914:230–236.


Targeting of the Retinal Pigment Epithelium (RPE) by Means ...

Documents