Imaging of the Optic Nerve and Retinal Nerve Fiber Layer: an Essential Part of Glaucoma Diagnosis and Monitoring

Glaucomatous damage specifically affects retinal ganglion cells (RGCs) and their axons and leads to progressive thinning of the retinal nerve fiber layer (RNFL) accompanied by structural changes within the optic nerve head (ONH). These changes typically appear as diffuse or focal enlargement of the optic cup corresponding to neuroretinal rim loss, displacement, barring and variation in caliber of the retinal vessels, optic disc hemorrhages, RNFL defects, and peripapillary atrophy. The detection of early glaucomatous optic nerve damage during a clinical examination is often challenging, even for experienced clinicians, because of a wide range of normal ONH appearances. Similarly, the interpretation of optic disc stereo-photographs, traditionally used to document the appearance of the ONH over time and to provide evidence of progression, is highly subjective, whereas poor quality images and differences in focus, exposure, magnification, and camera angle may often result in a false impression of progression.

The progressive loss of RGCs can lead to visual field (VF) defects. VF assessment routinely performed with standard automated perimetry (SAP) is subjective and prone to high short- and long-term inter-test variability. Because this variability can confound the assessment of VF changes over time, it is desirable to confirm these changes with objective test methods. Additionally, in many cases the loss of RGCs and resulting thinning of the RNFL and damage to the ONH have been shown to precede the onset of a glaucomatous VF defect.^{22; 48; 49; 57} Because glaucomatous damage is irreversible, early detection of structural changes is imperative for timely diagnosis of glaucoma and monitoring of glaucomatous progression.

Considerable improvements in ocular posterior segment imaging have been made in recent years. Imaging techniques such as optical coherence tomography (OCT), scanning laser polarimetry (SLP) and confocal scanning laser ophthalmoscopy (CSLO) rely on different properties of light to provide objective structural assessment of the RNFL, ONH and macula, thus assisting clinicians in the diagnosis of glaucoma and monitoring of its progression.

ONH and RNFL Imaging For Glaucoma Diagnosis

Optical Coherence Tomography

Based on the principle of low coherence interferometry, OCT provides cross-sectional visualization of ocular structures.^{26; 52; 53} Low-coherence light is aimed at a beam splitter, which splits the light and directs it to the retina and a reference mirror. The light reflected from the mirror then recombines with the light reflected from the retina creating an interference pattern caused by the altered magnitude and time delay of light as it encounters different optical reflectance across the depth of the tissue. Segmentation algorithms can be applied to the cross-sectional images to obtain retinal and RNFL thickness and ONH structural information. Stratus OCT (Carl Zeiss Meditec, Dublin, CA), the most commonly used time-domain OCT (TD-OCT) produces cross-sectional images with an axial resolution of 8 – 10µm and a transverse resolution of approximately 20µm. Peripapillary RNFL thickness measurements are obtained using a 3.4 mm diameter circular scan centered on the ONH. RNFL thickness is automatically determined and reported as an overall mean, by quadrants, and by clock hours. Quantitative information is also provided for the ONH structures and for total macular thickness.

TD-OCT RNFL and ONH measurements have been shown to discriminate well between healthy and glaucomatous eyes.^{4; 7; 39; 40; 45; 67} Mean RNFL thickness, inferior quadrant and superior quadrant RNFL thickness provide the best diagnostic accuracy. There is good measurement reproducibility for both diffuse and focal RNFL defects.^{3; 5; 6; 21; 47; 54}

Spectral-domain OCT (SD-OCT)

SD-OCT is a newer generation of the OCT technology offering several benefits over TD-OCT, such as enhanced resolution (3–6 µm axial resolution) and faster scanning (40–110 times faster with commercial SD-OCT systems).⁵¹ Unlike TD-OCT, which requires the use of a moving reference mirror to record the depth information of the reflections from the target tissue, SD-OCT captures this information in the frequency domain, enabling all the reflections included in one A-scan to be captured simultaneously.⁶⁴ Moreover, SD-OCT offers image registration and 3D rendering capabilities. These advantages result in improved measurement reproducibility compared with TD-OCT.^{18; 27; 33; 51} Although most studies showed that the glaucoma diagnostic ability of SD-OCT is similar to TD-OCT,^{9; 33; 51; 55; 61} SD-OCT has an improved capability in early-stage glaucoma detection.⁴⁶ Figure 1 shows an example of an early glaucomatous damage detected by SD-OCT.

Figure 1.

Early glaucomatous damage as detected by spectral-domain optical coherence tomography. Wedge shaped thinning of the retinal nerve fiber layer (RNFL; red zones in the RNFL Deviation Map) are evident in the superior and inferior regions in the right eye. RNFL Thickness profile also demonstrates localized thinning in the same regions. Quantitative analysis shows deviation from normal range in the overall, quadrant and clock-hour RNFL thickness analysis.

Scanning Laser Polarimetry

SLP (GDx, Carl Zeiss Meditec, Dublin, CA) determines the peripapillary RNFL thickness by measuring the amount of the retardation of polarized light, which is linearly correlated with the birefringent properties of the retina. As a result of the parallel orientation of the microtubules within the RGC axons, a change in the polarization of light, called retardation, occurs when light passes through the RNFL. This change can be quantified and is proportionate to the thickness of the RNFL.⁶³ The most recent commercially available iterations of this technology are named GDx VCC (Variable Corneal Compensator), GDx ECC (Enhanced Corneal Compensator), and GDx PRO. These devices provide individualized compensation for the birefringence of the media (mainly the cornea). The ECC version is an improvement over VCC resolving most of atypical retardation patterns (ARPs) that confound RNFL thickness measurement in a substantial subset of healthy and glaucomatous eyes. GDx provides reproducible measurement of RNFL thickness.^{13; 23; 63; 69} GDx VCC outperforms the earlier iterations of this technology that used fixed corneal compensation in its ability to discriminate between healthy and glaucomatous eyes.^{12; 20; 59; 62} GDx ECC performs better than VCC in the detection of early glaucoma.^{36; 56} GDx NFI (Nerve Fiber Index), a machine classifier parameter that combines several measurements, consistently offers the best diagnostic performance.^{14; 15; 50} However, this parameter is no longer available in most recent iterations of this technology and average TSNIT, quantifying the RNFL thickness, is the best diagnostic parameter among the existing parameters. Figure 2 shows an example of an early glaucomatous damage detected by the GDx PRO device.

Figure 2.

Scanning laser polarimetry imaging showing superior and inferior retinal nerve fiber layer (RNFL) atrophy in the right eye and inferior RNFL atrophy in the left eye (deviation map). Compared with age matched healthy controls, several retardation parameters are outside normal limits and marked by the red background (RNFL Summary Parameters table).

Confocal Scanning Laser Ophthalmoscopy

The CSLO (Heidelberg Retina Tomograph (HRT); Heidelberg Engineering, Heidelberg, Germany) uses a 670nm diode laser beam with a confocal detector device that scans the ONH and provides three-dimensional measurements of ONH topography. It then generates a number of stereometric parameters, such as rim area, cup area, rim volume, and cup-to-disc ratio. The device has good reproducibility^{10; 16; 30} and glaucoma discriminating ability,^{2; 41; 44; 65; 66} comparable to optic disc assessment by glaucoma experts.⁶⁵ In Ocular Hypertension Treatment Study participants, HRT was able to detect structural glaucomatous changes up to eight years before functional defects were seen on VF testing.⁷¹

The latest version of the CLSO, the HRT III, offers a large normative database as well as advanced analytical tools such as the Moorfields Regression Analysis (MRA)⁶⁶ and the Glaucoma Probability Score (GPS).⁵⁸ The MRA improves the diagnostic accuracy of the instrument by using the global and sectoral neuroretinal rim area adjusted for disc size and age. This method is highly capable of discriminating between healthy and glaucomatous eyes.^{42; 66} The GPS provides disease probability scores and minimizes operator error by relying on an automated approach to the optic disc classifying procedure. The discrimination ability of the GPS is similar to the MRA.⁸ An example of early glaucomatous damage detected by the HRT III is shown in Figure 3.

Figure 3.

Early glaucomatous damage as detected by Confocal Scanning Laser Ophthalmoscopy. Neuroretinal rim defect is marked in the temporal inferior region corresponding to an adjacent retinal nerve fiber layer defect (upper right). This region experienced a statistically significant deterioration from baseline as marked by the red region (upper left). Compared with healthy control data, several quantified parameters were outside the normal range.

Comparison of imaging technologies for glaucoma diagnosis

Studies comparing the glaucoma discriminating ability of TD-OCT, GDx and HRT^{19; 40; 70} demonstrated that the best parameters of all three instruments performed similarly. Subjective evaluation of the ONH by glaucoma experts was as good as the objective imaging modalities in discriminating glaucomatous and healthy eyes; however, this may not reflect common clinical practice as it has been shown that glaucoma experts perform better than general ophthalmologists.¹ Indeed, it has been recently demonstrated that the diagnostic ability of the three imaging techniques was better than subjective assessment of the ONH by general ophthalmologists.⁶⁰

SD-OCT may outperform SLP and CSLO in ability to diagnose glaucomatous damage. VF defects correlated better with RNFL thickness loss measured by SD-OCT compared to RNFL thinning as measured by SLP.²⁵ SD-OCT may have a higher sensitivity for glaucoma detection than HRT.³⁵

Monitoring of glaucoma progression

Glaucomatous progression typically occurs either as a continuous linear process where tissue and function are gradually affected or in a stepwise pattern where sudden damage caused by an acute event is followed by a period of minimal change that lasts until another acute event takes place. In some individuals these two scenarios may coexist or they can occur in different phases of the disease. As the exact mode of progression in a given subject cannot be easily predicted, the assessment of glaucomatous progression requires clinical judgment as well as event- and trend-based analyses. In trend analysis, regression analysis of a dependent variable (e.g. RNFL thickness) on serial measurements provides progression rate over time. In event analysis, progression occurs when a follow-up measurement exceeds a pre-established threshold for change from baseline.

In clinical practice automated perimetry has been the standard for detecting glaucoma progression. In many subjects, however, structural damage precedes VF changes or occurs without simultaneous progressive VF changes.^{43; 48; 57} This creates the need for tools capable of reliable and objective evaluation of progressive structural changes. Longitudinal assessment of glaucoma with imaging devices poses a significant challenge because of the rapidly evolving technology and resulting frequent software and hardware changes. Moreover, because of the lack of commonly acceptable reference standard that can be used to indicate progressive glaucomatous change, it is difficult to determine whether progression detected by an imaging device in the absence of VF loss reflects structural changes preceding functional loss as measured by SAP or is a false positive.

Most OCT progression studies use TD-OCT because of the longer follow up period.^{68, 31; 32; 37} In a study on 64 glaucomatous and glaucoma suspect eyes followed for a mean period of 4.7 years, progression events defined by OCT occurred more frequently than progression events defined by VF.⁶⁸ The OCT progression studies in which progression was defined based on red-free fundus photographs demonstrated that analyzing both the mean and sectoral RNFL thicknesses is important in maximizing the detection of progression.^{31; 32} The inferotemporal (7 o'clock) sector and inferior quadrant RNFL thickness are most predictive of progression.^{24; 28; 32; 37}

The advantages offered by SD-OCT result in improved intra-visit and inter-visit measurement reproducibility, indicating this instrument’s potential for detecting early progression. In a recently published longitudinal study comparing SD-OCT with TDOCT, out of 128 glaucomatous eyes that were followed for a minimum of two years, 19 and 4 eyes were identified as progressing with SD-OCT and TD-OCT, respectively.³⁴ Figure 4 shows an example of glaucomatous progression detected by SD-OCT.

Figure 4.

Spectral-domain optical coherence tomography retinal nerve fiber layer (RNFL) Guided Progression Analysis. Likely progression (marked in red) is identified in the inferotemporal and superotemporal regions on the RNFL Thickness Map and RNFL Thickness Profiles Progression. Average RNFL Thickness Progression shows likely progression in the overall, inferior and superior sector thickness. Trend lines are drawn and rates of change are provided all of which show statistically significant rate of RNFL thinning.

SLP derived structural measurements have a higher sensitivity for progression detection than SAP.¹¹ In a recent study evaluating progression with GDx-ECC, rates of RNFL loss were significantly greater in eyes that showed evidence of glaucoma progression based on SAP and/or optic disc stereophotographs compared to eyes that remained stable.³⁸ The GDx-ECC version of this technology performed significantly better than the VCC version for detection of change, suggesting that it could improve longitudinal evaluation of the RNFL. Figure 5 shows an example of glaucomatous progression detected by GDx.

Figure 5.

Scanning laser polarimetry Guided Progression Analysis. The Image Progression Map shows progression in the inferotemporal and superotemporal regions. TSNIT Progression Graph shows progression in the superior region. Summary Parameter Charts show a significant decrease in TSNIT average, superior and inferior RNFL thicknesses.

Several longitudinal studies compare HRT with optic disc stereophotography and SAP in glaucomatous patients. In these studies, progression identified by HRT’s topographic change analysis (TCA) was more frequent than progression identified with expert stereophotographic assessment of ONH and progression identified by SAP.^{17; 29} There was a poor agreement, however, among the three techniques. The discrepancy was attributed to the inability of HRT change analysis to detect features such as splinter hemorrhages or defects in nerve fiber layer. HRT can detect small topographical changes that are otherwise not easily appreciated. Figure 6 shows an example of glaucomatous progression detected by HRT.

Figure 6.

Scanning laser ophthamoscopy Topographical Change Analysis (TCA). A surface height depression (red pixels on follow-up scans) can be seen initially in the inferotemporal region. Progressive wedge shaped enlargement of the depressed area is seen on the follow-up images.

Conclusion

Imaging devices play an essential role in the diagnosis of glaucoma by providing a set of objective quantitative measurements and statistical classifications using comparisons to normative data. They cannot, however, reliably detect certain abnormal features such as an optic disc hemorrhage or disc pallor and therefore should not replace a clinical examination. Instead, they provide important complementary information intended to assist the clinician in the diagnostic process. As these technologies undergo constant evolution, their capabilities continue to improve, both for disease detection and identification of progression. None of the three main imaging modalities has been reliably demonstrated to be superior to another, although recently published studies using SD-OCT indicate that this technology may outperform SLP and CSLO in its ability to detect glaucomatous damage. Imaging-derived measures of progression have been shown to be more sensitive to change than SAP but, without the reference measure of progression, the reliability of these measures remains unknown. Due to the temporal dissociation between structure and function and the fact that glaucoma is a slowly progressive disease, longer follow-up periods are needed to establish whether structural changes identified with these imaging technologies can predict the subsequent development of VF loss. Because imaging may falsely identify glaucoma and its progression, clinical management decisions should always be based on a combination of structural and functional measures and the results of a clinical examination.