Ординатура / Офтальмология / Английские материалы / Essentials in Ophthalmology Glaucoma_Grehn, Stamper_2008

Nisha R. Acharya

Francis I. Proctor Foundation

University of California

San Francisco

95 Kirkham St., San Francisco

CA 94143, USA

Iqbal Ike K. Ahmed

University of Toronto

Toronto

Ontario, Canada

Anne Louise Coleman

Jules Stein Eye Institute/UCLA

100 Stein Plaza

Los Angeles, CA 90095

USA

M. Francesca Cordeiro

Glaucoma & Retinal Degeneration

Research Group

UCL Institute of Ophthalmology

Bath Street

London EC1V 9EL, UK

Amish B. Doshi

Hamilton Glaucoma Center

Department of Ophthalmology

University of California

San Diego, CA

USA

Paul Foster

Department of Epidemiology &

International Eye Health

UCL Institute of Ophthalmology

11-43 Bath Street

London EC1V 9EL, UK

David F. Garway-Heath

Moorfields Eye Hospital and UCL

Institute of Ophthalmology

NIHR Biomedical Research Centre

162 City Road

London, UK

Annette Giangiacomo

CB 7040, 5109 Bioinformatics Building

Department of Ophthalmology

University of North Carolina-Chapel Hill

Chapel Hill, NC 27599-7040 USA

Franz Grehn

University Eye Hospital Wuerzburg

Josef Schneider Str. 11

97080 Wuerzburg

Germany

Li Guo

Glaucoma & Retinal Degeneration

Research Group, UCL

Institute of Ophthalmology

Bath Street

London EC1V 9EL, UK

John H.K. Liu

Hamilton Glaucoma Center

Department of Ophthalmology

University of California

San Diego

CA, USA

Sancy Low

Glaucoma Service

Moorfields Eye Hospital

London, UK

Department of Epidemiology and

International Eye Health

UCL Institute of Ophthalmology

Bath Street, EC1V 9EL

London, UK

Efstratios Mendrinos

Department of Ophthalmology

Glaucoma Unit

Geneva University Hospitals

1211 Geneva 14

Switzerland

Agnieszka G. Nagpal

Francis I. Proctor Foundation

University of California, San Francisco

95 Kirkham St.

San Francisco, CA 94143

USA

Marc Schargus

University Eye Hospital Wuerzburg

Josef Schneider Str. 11

97080 Wuerzburg

Germany

Tarek Shaarawy

Glaucoma Unit

Department of Ophthalmology

Geneva University Hospitals

Alcide-Jentzer 22

1211 Geneva 14

Switzerland

Tarun Sharma

Glaucoma Service

Moorfields Eye Hospital

London, UK

Nicholas G. Strouthidis

Moorfields Eye Hospital and

UCL Institute of Ophthalmology

NIHR Biomedical Research Centre

162 City Road

London, UK

Diamond Y. Tam

University of Toronto

Toronto

Ontario, Canada

Robert N. Weinreb

Department of Ophthalmology

University of California

9500 Gilman Drive

La Jolla, CA 92093 USA

Nick Wood

Glaucoma & Retinal Degeneration

Research Group

UCL Institute of Ophthalmology

Bath Street

EC1V 9EL London, UK

■ Confocal scanning laser ophthalmoscopy (cSLO): A confocal imaging system which uses a fine confocal aperture to limit the light detected to that from the focal plane, and therefore achieves high lateral

1resolutions

■Adaptive optics (AO): An adaption which uses a patterned guide laser to sense errors in the optics of the eye and a deformable mirror to correct for them in real time

■Retrograde labelling of RGCs via direct application of dyes has enabled the analysis of ganglion cell number and morphology in numerous studies with animal models

■RGC-specific fluorescent protein expression has been developed in a number of mouse lines to enable RGC identification and subtype study

■The detection of apoptosing retinal cells (DARC) uses an injection of fluorescently labelled annexin-5 which binds to the membrane of apoptosing cells to act as a marker for RGC disease

Summary for the Clinician

■Glaucoma is the leading cause of irreversible blindness if left untreated, and current methods will only detect it when significant damage has already been done

■RGCs are the key cells implicated, and observing them could lead to effective treatment and monitoring regimens

1.3 The Imaging Techniques

The imaging of cells in living systems poses numerous problems, as (unlike histology) it involves direct exposure of living tissue, and even relatively innocuous staining compounds bind to cell constituents and therefore may interfere with cellular function. Intrinsic cellular properties are therefore sought that allow them to be resolved from the surroundings. The RGCs in particular have proven a challenge to image, but modern techniques taken from other fields such as cell biology and cosmology are beginning to yield some insight into their morphology and behaviour in vivo.

Many techniques have recently been developed to assess the RNFL thickness, as its thinning is associated with glaucomatous progression [10]. In real terms, this thinning process represents the large-scale loss of the RGC axons. However, higher resolutions are needed to gain access to individual cell bodies, and here we discuss the most current methods and some of the promising directions the research is taking.

1.3.1Scanning Laser Polarimetry

First reported by Weinreb et al. in 1990 [11], the basic layout of this can be seen in Fig. 1.1. This is based on the linear relationship between the birefringence and thickness of the RNFL. Birefringence is a quality of highly ordered optical media such that they exhibit polarising properties and refractive indices that are dependent on the polarisation of the incident light. This can be detected by a system with polarisation-sensitive detectors. The parallel microtubule structures in the RNFL cause birefringence and the degree of birefringence is dependent on the tissue thickness. This measure has been shown to be sensitive and specific and, unlike the cSLO technology (Heidelberg Retinal Tomography (HRTIII), Heidelberg Engineering Vista, CA, USA), it has the advantage of not requiring the operator to provide reference points [12]. The cornea had previously been a problem in SLP imaging, as it also has a degree of birefringence. The current incarnations of the commercially available machine (the GDx, Carl Zeiss Meditec, Inc., Dublin, CA, USA) have overcome this [13] with a variable corneal compensator (VCC-SLP) which uses the macular as a reference point non-birefringent to gain a measure of the corneal birefringence. This was followed by the enhanced variable compensator (ECCSLP), which avoids problems with low-quality images [14] by using software correction. The machine is still limited to measuring the RNFL thickness though, and is therefore not as versatile as the other devices in terms of RGC cell body assessment.

Summary for the Clinician

■SLP uses the polarisation change imparted on incident light by the RNFL to measure its thickness

■VCC and ECC have been developed to counter problems with corneal birefringence

■The machine is limited to RNFL thickness analysis

1.3.2High-Resolution Reflectance Imaging

This uses a high-quality, high-speed CCD attached to a fundus camera with a method for the illumination of the retina in time with image detection. The sensitivity of the camera allows for very accurate measurement of reflectivity changes. The stimulation of nervous tissue has been shown to cause reflectivity changes [15–17]. Such changes are most likely due to changes in membrane reflective index or morphology and can be detected with the sensitive camera to give an indication

of the function of the cells. When used in the eye, imaging the inner layers can give reflectivity changes representing the activation of RGCs [15], and may be able to give a measure of their functional health. However, the absolute quantification of RGCs is not possible. A commercial version, the Retinal Functional Imager (RFI, Optical Imaging, Rehovot, Israel), promises to allow blood flow velocity calculations using ratio comparison of images, and will be able to rapidly change a filter wheel in time with the image acquisition to gain high temporal resolution data on wavelength-depend- ent reflectivity changes [18].

Summary for the Clinician

■Reflectivity changes can be measured with high acuity and temporal accuracy using a fundus camera with a high-quality CCD and light source

■These machines are capable of detecting functional reflectance changes and measuring blood velocity

1.3.3Optical Coherence Tomography

Based on a technique long established in other fields, OCT was developed to visualise the retina in 1991 [19]. In the last few years, OCT has become an increasingly accepted method for assessing the thickness of the RNFL [20] due to its reproducibility and accuracy [21]. This has fuelled a series of rapid advances, and current research is demonstrating submicron resolutions [22, 23], functional imaging [16, 17] and 3D imaging [24]. Imaging of the RGC layer however, is difficult due to its weakly backscattering nature and low-contrast edges, and the technology has some way to go before individual cells become accessible in vivo.

The basic form is shown diagrammatically in Fig. 1.2. This system is based on low-coherence interferometry, the principle that light split from one beam into two different paths which are then recombined will create interference patterns that depend on the phase difference introduced by the variance in the path length. When the beams are recombined in phase they will do so constructively, but they destructively combine when they are half a coherence length out of phase. The coherence length is the distance

over which a wave can combine to form interference patterns. Therefore, the smaller the coherence length, the more certain the distance measured by interference will be. The coherence length is inversely proportional to the coherence of the light source. Images can be constructed from a number of ultrasound-like A scans to give a 2D reflectivity profile, or B scan. A 3D image is possible if sufficient B scans can be taken before involuntary movement affects slice alignment. The scanning time in any particular axis is a major time and complexity limitation of the devices, so a number of ingenious methods have been adopted to reduce total imaging time and increase resolution.

The first OCT incarnations used a moving reference arm to obtain axial information. These are called time domain OCTs (TD-OCT), as the axial information is obtained over the time taken to move the reference arm. Scanning speed is thus limited by the movement of the motors controlling the reference mirror. The time taken to make a series of scans sufficient to construct a 3D image is preclusive, as involuntary eye movement will mar the results. There are a number of clinically available machines, such as the Stratus (Carl Zeiss), which have proven efficient at disease detection [21, 25–28], but with maximum resolutions of 10 microns they are unable to clearly resolve the RGC layer.

Ultrahigh resolution OCT (UHR-OCT) is a time domain system that uses a wide-band light source which gives a much smaller coherence length than conventional OCT. This improves the axial resolution and has given three-micron resolutions in the retina [29]. The increase in resolving power has enabled the imaging in vivo of many retinal layers, including the RGC layer (although not individual RGCs), which is impossible to image with conventional TD-OCT [29–33]. While their expense has traditionally been preclusive, as they rely on costly broadband lasers, developments in optical technology have now begun to offer alternatives, such as super luminescent diodes (SLDs) [30, 34] or xenon arc lamps [22]. The resulting reduction in cost is bringing UHR-OCTs to the clinic.

Ultrahigh-speed spectral domain OCT (SD-OCT) is a faster, more sensitive method for obtaining axial information which measures the interference for each wavelength in a broadband light source in order to obtain full axial scans without moving the reference or sample arm [35–37]. This is because the distance at which interference occurs will vary with wavelength. The detector can be replaced by a spectrophotometer to measure all the broadband spectra at the same time. Alternatively, the light source can be replaced with a swept source, and the same photometer as used in TC-OCT can be used.

Because the system is not limited by the rate of sample mirror movement, the scans take less time and 3D imaging as well as video-rate 2D sectioning becomes possible [30, 36]. The system is also called Fourier domain OCT (FD-OCT), as the signals obtained are related to those of TD-OCT by the Fourier transform. Devices using FDOCT have recently been made commercially available, and are taking full advantage of the high image capture rates [40,000 A-scan/s with the Spectralis (Heidelberg Engineering) and Cirrus (Carl Zeiss)] to make 3D images of the diseased eye available in the clinic [38, 39].

En face OCT has been used in a number of studies to attempt to overcome the imaging speed limitations of TDOCT. Two different methods have been developed to provide an en face image in real time. Full-face methods have been used to create three-dimensional reflectivity images at resolutions approaching those of confocal machines [22, 23, 34, 40–42]. These machines use high numerical aperture (NA) lenses on the reference and sample arm and a charge-coupled device (CCD) to capture full-face (x–y) images. As there is no lateral scanning, transverse cross-sections can be captured in real time, and because of the high-NA lenses, the submicron resolution is sufficient to access cellular detail. There are limitations to the systems though, and many modifications need to be made before they can be used for disease detection. The small imaging volume and requirement for the sample to be moved for axial scanning make it unsuitable for in vivo studies. Additionally, the use of high-NA lenses causes aberrations when attempting deep tissue penetration [22]. Flying spot systems use a fast transverse raster scanning technique combined with slower axial scanning. This allows the construction of en face scans in real time [34]. The main advantage of this system is the ease with which it can be combined with an SLO [34, 41, 42]. This overcomes the axial resolution limitation of SLO as well as the lateral limitation of traditional OCT. The scanning speed is still limited by the slowest moving part though (the axial movement systems in this case).

The OCT has also been shown capable of birefringence measurements if a polarimeter is combined with the light path [35]. This should allow yet more accurate measurements of RNFL thickness, and even allows a measure of birefringence with depth, which SLP is unable to achieve. This may give an indicator of disease-induced structural losses in specific regions of the nerve fibre layer. One advantage of the OCT polarimeter over the SLP is that it can use the surface of the RNFL as a reference point and is therefore not susceptible to variations in the corneal birefringence.

Functional OCT (fOCT) is becoming a reality as more studies are being made of the changes in reflectivity

profiles of nervous tissue with stimulation [15–17]. The fOCT can enable the examination of functional reflectivity changes through all layers of the retina over time [17], and has shown changes in the RGC layer after light stimulation. As the frame rate of the OCT becomes greater, it may become possible to image 3D volumes with sufficient temporal resolution to form maps of retinal function in vivo.

A full review of OCT is beyond the scope of this book chapter, but we would recommend [41] for further reading.

Summary for the Clinician

■OCT has been around for over a decade and is increasingly used for RNFL analysis

■It is based on low-coherence interferometry

■Time domain OCT uses a moving reference arm for image acquisition

■Ultrahigh resolution OCT uses a broadband light source to increase resolution

■Spectral domain and Fourier domain OCT uses the spectrally varying coherence length of light to drastically increase scan rates

■En face OCT uses different methods to obtain full-face images in real time

■OCT machines can also be made to measure function dependent reflectivity changes and birefringence

■As yet, due to the weakly backscattering nature and low-contrast edges of cells, individual RGCs are not able to be seen with this technology

1.3.4Confocal Scanning Laser Ophthalmoscopy

Currently the most popular and well tested of the imaging modalities discussed, the cSLO, has survived the fastmoving field well due to its versatility. It is based on the principle of confocal microscopy (Fig. 1.3): namely that if light is focussed onto a spot in the focal plane by passing it through a fine pinhole, and the returning light is passed back through the same pinhole, then the action of the fine aperture is to exclude scattered light and light reflected from planes above or below the focal plane. This allows much higher axial and lateral resolution than achieved with conventional light microscopy, such as with a fundus camera. Because they are based on confocal systems, cSLO are inherently suited to viewing excited fluorophores. This capability has been used to detect the

apoptosis of individual RGCs [3, 8, 9, 43] as well as to longitudinally view fluorescent markers expressed in the ganglion cells of mice in vivo [44].

Their ability to create 3D topographies, a long-estab- lished technique in the clinical assessment of eye disease, can even be used to make indirect assessments of the RNFL thickness [12].

Machines that combine FD-OCT and SLO may overcome the axial limitations of SLO and the lateral limitations of OCT, so this technology may allow the accurate positioning of fluorescent markers in both axial and lateral space. There are now commercially available combination machines, such as the Cirrus and the Spectralis.

Summary for the Clinician

■cSLO uses fine apertures to limit the light detected to that from the focal plane

■Clinical machines can indirectly measure RNFL thickness

■Modern machines combine cSLO with OCT to overcome the axial resolution limitations of cSLO and the transverse resolution limitations of OCT

1.3.5Adaptive Optics

Many changes have been made to the current imaging systems over the years, producing a number of improvements in the optics and cameras, and a technique learned from a branch of physics is proving very useful in both OCT and SLO.

AO was originally developed for cosmologists to eliminate the speckle produced by the passage of light through the atmosphere when viewing astronomical objects from the Earth. It is based on the idea that if a reference beam consisting of a 2D array of equidistant spots is passed along the same light path as the measurement beam but does not collect information from the object measured, then deformations from the reference pattern must be due to interference in the optical media. These deformations can then be corrected in real time by reflecting both the reference and measured beams off a computer-control- led deformable mirror before passing the measurement beam to the detector and the corrected reference beam to the Hartman–Shack wavefront sensor for further corrections to the mirror. When imaging the eye, the pattern beam is reflected from a point on the retina to prevent the retinal surface scattering the grid pattern.

AO was introduced in ophthalmology to reduce the effect of aberrations in the optical media of the eye and so increase the signal-to-noise ratio (SNR) at high resolutions. This has improved acuity to the extent where direct observation of the cone cells has been performed using AO-SLO via their intrinsic autofluorescence alone.

It has also been applied to OCT [30, 45], where the use of AO has improved the transverse resolution to the extent that individual nerve fibre bundles have been observed with a FD-OCT in humans in vivo [38].

Recently, AO has been used to improve in vivo imaging quality in mouse eye, where AO was incorporated into a biomicroscope to overcome the presence of aberrations in the rodent eye [46]. With improved resolution, it is possible to image fluorescently labelled capillary vessels and dendrites of microglial cells in mice.

Summary for the Clinician

■AO uses a patterned reference beam to detect aberrations in the optical medium

■Imperfections can be corrected in real time using a deformable mirror

■This improves the SNR, allowing fine detail to be seen

1.4Applications to RGC Imaging

Until imaging technology achieves higher resolutions, RGC-specific contrast-enhancing agents will have to be used. Intrinsic methods may never be able to detect subtle specific gene expression changes in subtypes of RGCs without using extrinsic agents, which can overcome the resolution limitations by marking features with obvious fluorescent cues. Because of this, the development of a number of contrast-enhancing agents has proven invaluable in glaucoma research.

1.4.1Retrograde Labelling

Carbocyanine dyes such as DiAsp (4-[4-didecylaminos- tyryl]-N-methyl-pyridinium iodide), DiI (1,1′-dioctade- cyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate), DiO (3,3′-dioctadecyloxacarbocyanine perchlorate) and Fluorogold are lipophilic dyes. Following application, they are inserted into the outer leaflet of the cell membrane and then, as part of lipid turnover, are transported within the cells [47]. Alternatively, hydrophilic dyes such as rhodamine dextran, which are actively transported throughout the cell, can be used. If applied high up in the visual pathway, these fluorescent dyes will pass down the length of the RGC axons into the cell bodies in the retina. This method has been the traditional way of labelling the RGC population of the model organism retina [47, 48, 48–50], and has been used in numerous studies of glaucoma [5, 51–53].

The majority of experiments performed with retrograde dyes have been ex vivo, but retrograde labelling is also showing results with in vivo applications [5, 51, 52]. There are a number of techniques for application, such as direct injection into the axonal innervating centres of the brain (the superior colliculi), or placement of crystals onto the optic nerve.

A very recent demonstration has shown that it is possible in primate retina to view the characteristics and morphology of rhodamine-labelled RGCs (Gray, IOVS 2008). Figure 1.4 shows how this is made possible by applying AO techniques to a cSLO image.

All methods have strong limitations though, as they have the distinct potential to cause damage to the cells and are very invasive. It is because of this that this method cannot be used to examine human cells in vivo. It has produced a huge number of studies in animal models though, and a similar, safer technique in humans would be a very powerful tool.

Summary for the Clinician

■Retrograde labelling uses lipophilic dyes applied in the visual pathway to label RGCs

■It has traditionally been used with histological studies, but in vivo research is now producing good results

■It cannot be used in humans because of the application methods, but it can be used with animal models to produce clinically relevant data

1.4.2RGC-Specific Fluorescent Protein Expression

Modern genetic techniques allow us great control over gene expression in a number of model species and provide us with tools such as green fluorescent protein (GFP) expression systems for identifying cells manufacturing particular proteins. The study of RGCs has produced a number of specific markers, such as Thy1, c-kit and Brn3b in rodents, which are specifically expressed in RGCs over other neurons in the retina [5, 6, 45]. A combination of a Thy1 promoter with introns specific to RGCs and a GFP marker has been used to contrast-enhance RGCs and even distinguish between subtypes in mice [7]. This may prove a good resource to test for RGC subtype-specific disease susceptibilities, as with advances in imaging technology the mouse eye is becoming available for noninvasive studies [43] which allow the longitudinal assessment of disease models [54]. However, this technique is limited to animal studies, as gene insertion is not a viable technique for human studies. Much transferable data can be gained though, which will greatly aid human studies.

Summary for the Clinician

■Genetic techniques have produced animals which specifically express fluorescent markers in RGCs

■These animal models can be used to assess the affects of disease on RGCs and even subpopulations of ganglion cells

1.4.3The Detection of Apoptosing Retinal Cells (DARC)

In moving towards clinical applications, disease detection methods must be refined from the corresponding research techniques. The reflective response to light would give a general intrinsic signal of RGC function, but variability must be reduced and the RGC layer assessed separately from the rest of the retina before it specifically indicates ganglion cell health. Good extrinsic markers for retinal health are therefore needed to make up for shortcomings in the imaging techniques.

DARC has recently been developed as a novel, noninvasive real-time imaging technique for the visualisation of individual RGC apoptosis in the living animal [3]. Apoptosis is the process by which cells undergo controlled destruction following an injury. The technique involves the application of fluorescently labelled annexin-5 to detect apoptosing cells using a fluorescence cSLO such

as the HRA. All of the fluorescent spots across a large section of the retina can then be counted to give a quantitative measure of the level of RGC health. Annexin-5 preferentially binds to phosphatidylserine (PS) that translocates from the inner to the outer plasma membrane during apoptosis. This step occurs much earlier than latestage markers such as DNA fragmentation, which can be detected by methods such as TUNEL, as the PS is used by the cell to indicate the start of the apoptosis cascade.

DARC promises to deliver a method for monitoring glaucomatous disease progression and detecting glaucoma at an early stage before large-scale RGC loss has occurred. Using DARC, it has become possible to image changes occurring in RGC apoptosis over hours, days, and months in glaucoma-related animal models in vivo for the first time.

Currently, there is no adequate method for assessing neuroprotection in glaucoma. This drawback is reflected in the expensive six-year period of follow-up for the only