Clinical Utility of Optical Coherence Tomography in Glaucoma.

Optical coherence tomography (OCT) has established itself as the dominant imaging modality in the management of glaucoma and retinal diseases, providing high-resolution visualization of ocular microstructures and objective quantification of tissue thickness and change. This article reviews the history of OCT imaging with a specific focus on glaucoma. We examine the clinical utility of OCT with respect to diagnosis and progression monitoring, with additional emphasis on advances in OCT technology that continue to facilitate glaucoma research and inform clinical management strategies.

Glaucoma is a multifactorial, progressive, degenerative optic neuropathy and is the second most common cause of blindness worldwide.[1] The disease is characterized by the death of retinal ganglion cells (RGCs) and their axons and by associated morphologic changes within the optic nerve and retinal nerve fiber layer (RNFL).[2-7] Progressive neuroretinal rim thinning and excavation of the optic nerve head (ONH) are consistent findings. Although most glaucoma types progress slowly, the disease can lead to blindness without treatment.[1,8] With treatment, glaucomatous progression can often be slowed or stopped. Accurate and early detection of glaucoma, therefore, is critical to successful management.

Various imaging modalities have increased and decreased in popularity as adjunctive technologies for the diagnosis and progression monitoring of glaucoma.[9] Optical coherence tomography (OCT) has become the technology of choice. Optical coherence tomography was first demonstrated in 1991[10] as an application of low-coherence interferometry.[11] Enabling noninvasive, high-resolution cross-sectional imaging of the retina in vivo, OCT's clinical utility for glaucoma was quickly realized.[12,13] Optical coherence tomography became commercially available in 1996 after scanning patterns with reproducible measurements were implemented by industry.[14] Optical coherence tomography has since changed the paradigm of assessment of the retina and revolutionized the management and diagnosis of glaucoma, allowing for objective and quantitative evaluation of neural structures affected by the disease, such as the macula and its individual layers, RNFL, and ONH.[15-23]

Optical coherence tomography technology has advanced since it was first applied to the eye and continues to rapidly evolve. Hardware advances in commercial systems improved resolution and increased scanning speeds. Previously available OCT instruments used a technique referred to as time-domain OCT (TD-OCT), which encoded the location reflections in the time information and related the location of the reflection to the position of the moving reference mirror, could obtain images of the fundus, discriminate glaucomatous eyes from normal, and detect change over time. However, this technology was limited by slow scan acquisition times and two-dimensional imaging.[24-33] The introduction of spectral-domain OCT (SD-OCT), which instead acquired all information within a single axial scan simultaneously through the tissue by evaluating the frequency spectrum of the interference between the stationary reference mirror and reflected light, increased reproducibility and accuracy in quantifying glaucomatous damage by further improving scan density and resolution and reducing imaging artifacts and scan acquisition time.[17,34-43] One of OCT's main strengths is its unparalleled high axial image resolutions. Previous TD-OCT B-scans had an axial resolution of approximately 10 μm, whereas the introduction of typical commercially available SD-OCT instruments improved resolution to approximately 5 μm axially with broad bandwidths at near infrared wavelengths. This greatly decreased the need for interpolation compared with TD-OCT.

Although SD-OCT significantly increased signal-to-noise ratio and decreased motion artifacts compared with TD-OCT, both are prone to image artifacts. These artifacts include speckle noise, segmentation and alignment errors, low signal quality, software problems, and media opacities, such as reduced corneal clarity or cataracts. Fixational eye movements, such as ocular tremor, ocular drift, and microsaccades,[44] can further reduce image quality. Microsaccades can instantaneously move the fixated point an average amplitude of 30 arcminutes,[45] causing severe motion artifacts and unreliable measurements. Although SD-OCT devices still require a trained imaging technician to minimize artifacts, it is less operator dependent than TD-OCT devices.[46]

RNFL thickness was the OCT parameter most often used in glaucoma assessment and provided objective, quantitative measurements of RNFL thickness,[10] but the addition of new parameters on commercial devices, such as those from superficial and deep structures of the ONH and macula, improved clinical utility of OCT. Scanning patterns were also developed that could deliver three-dimensional (3D) data. Spectral-domain OCT has since established itself as the dominant imaging modality in the management of glaucoma, although newer technologies are on the horizon, as written below. Today, SD-OCT instruments are available commercially from multiple manufacturers.[47] Resolution, scan acquisition rates, and measurements across these devices are generally not interchangeable, but their ability to detect glaucoma is very similar.[16,48,49]

Swept-source OCT (SS-OCT), a newer generation of OCT, has recently been commercially introduced.[50] Swept-source OCT uses a longer wavelength (generally 1050 nm) compared with SD-OCT (840 nm).[51] Swept-source OCT can evaluate RNFL and macular thickness, but can also more clearly image deeper ocular structures such as the choroid and lamina cribrosa (LC) in patients.[51-54] With SD-OCT, it is challenging to image deep structures due to relatively poor wavelength penetrance and decreasing sensitivity and resolution with increasing depth.[55] Repeatable and automated methods of quantification and visualization of LC, however, have been developed in SD-OCT augmented with techniques for correcting for optical aberrations in the eye[56] or with enhanced depth imaging,[57-60] which simply involves moving the SD-OCT device closer to the eye, allowing use of the image on the side of the zero delay opposite from the one utilized conventionally, with consequent improved signal from deeper structures rather than more anterior ones.[61-63] Swept-source OCT is able to scan at higher speeds and can acquire high-quality wide-angle scans that contain a large area of the posterior pole, including both the optic disc and macula.

The advent of SS-OCT systems also considerably improved the visualization of the anterior chamber (AC) structures, permitting objective quantitative analysis of the AC angle and facilitating comprehensive 3D assessment of the angle. Anterior segment OCT has improved with faster scanning speeds and newer algorithms for measurement of AC parameters and is generally more sensitive in detecting angle-closure when compared to gonioscopy.[64] Swept-source OCT affords high repeatability in terms of anterior chamber angle width measurements, such as angle opening distance, trabecular iris space area, and trabecular iris angle[65] and is able to capture extremely high resolution images of the AC. A consistent measurement of iris volume and the area of peripheral anterior synechiae is now possible with SS-OCT, whereas prior generations of OCT could not reliably assess these AC parameters.[66-69] Optical coherence tomography for evaluation of the anterior segment continues to evolve and will likely have an important role in the diagnosis and management of glaucoma patients.[70]

The application of OCT has also recently been extended to angiography and blood flow measurement. Techniques to perform OCT angiography recently became commercially available,[71] and ongoing studies are exploring the link between blood flow and glaucoma. Optical coherence tomography angiography offers a repeatable,[72] high-resolution, 3D quantitative evaluation of retinal vascular abnormalities in vivo and is a promising alternative to dye-based angiography, avoiding the dye injection-related complications. Reduced retinal perfusion in the ONH and peripapillary retina has been observed in glaucomatous eyes.[72-75] Whether decreased ocular blood flow in the ONH is the cause or the result of glaucoma progression remains unresolved.[71]

These methods are also not without limitations. The cross-sectional angiograms of OCT angiography devices, for example, often show projection artifacts[76,77] due to fluctuating shadows from the blood in the inner retinal vessels. Although these artifacts may be accounted for and removed, shadows from the choriocapillaris often obscure visualization of deeper choroidal vessels. Furthermore, OCT angiography may be unable to detect extremely slow blood flow, as is present in some pathologic conditions. Doppler OCT, although able to detect to motion parallel to the OCT beam, has a limited ability to visualize motion predominantly perpendicular to the probe, such as is the case in retinal and choroidal circulations. Although hemodynamic parameters may be useful in diagnosis[78] and management of glaucoma, its true clinical utility remains to be determined.

Diagnosis of Glaucoma

Glaucomatous structural damage often precedes vision loss.[24,79-84] Although diagnosis of moderate to severe cases of glaucoma is relatively straightforward, with diagnoses confirmed based on the presence of typical visual field (VF) defects on standard automated perimetry (SAP) and corresponding signs of glaucomatous ONH damage, the disease typically remains asymptomatic in the early stages. Standard automated perimetry has been widely used for diagnosis, staging, and monitoring of glaucoma, but is only likely to detect functional deficits after at least 20%–40% of RGCs have been lost.[3,24,78-80] Furthermore, visual field testing is often variable, and diagnosis may require repeated testing.[85,86] Identification of early glaucomatous structural damage, such as structural remodeling of the ONH and inner retinal layers, is essential for early diagnosis, management, and prevention of vision loss.[87-89]

Clinical assessment using multiple parameters, including peripapillary RNFL, ONH, and macular parameters, has proven useful, not only for management and diagnosing glaucoma at various levels of severity, but for evaluating risk in glaucoma suspects.[90] Although the use of multiple parameters could increase false-positive results, structural damage may be present in one parameter and not the other, and thus it is helpful to have information from the macula, ONH, and RNFL in glaucoma diagnosis.[78,84]

Current SD-OCT RNFL thickness parameters alone, however, have good diagnostic accuracy and help clinicians in determining severity stages and differentiating normal from glaucomatous eyes in the early stages.[40,48] Retinal nerve fiber layer parameters most extensively researched include: global average circumpapillary RNFL thickness (average of thickness measurements in the circumpapillary circle centered on the ONH), thickness deviation map, and thickness parameters measured by quadrants or clock-hour sectors. In general, average circumpapillary RNFL thickness and inferior sector RNFL thicknesses are the OCT parameters with the best diagnostic accuracy, with superior quadrant thickness values following in terms of sensitivity.[15,20,21,23] This agrees with prior studies showing superior and inferior areas of the optic nerve most commonly affected glaucoma[91-93] (Fig. 1). The diagnostic accuracy of SS-OCT circumpapillary RNFL parameters are similar to those of SD-OCT.[94] For detection of glaucomatous damage, the SD-OCT RNFL parameters have sensitivities ranging from 60% to 98% and specificities ranging from 80% to 95%.[15,16,18,20,21,23] Diagnostic performance decreases, however, for detection of early disease to 48% to 77% at the same specificity range for patients with minimal visual field losses.[95,96] Recent evidence shows, however, that even before the appearance of any VF defects on SAP, RNFL average thickness parameters could detect glaucomatous damage: at 95% specificity, up to 35% of eyes had abnormal thickness values 4 years prior to detectable VF loss and 19% had abnormal values 8 years prior.[84] The reproducibility of current SD-OCT RNFL thickness parameters is excellent, with global average RNFL thickness generally being the most reproducible.[97]

Figure 1

Retinal nerve fiber layer analysis from Spectralis-OCT (Heidelberg Engineering, Heidelberg, Germany) demonstrating glaucomatous damage. The RNFL evaluation in the right eye (OD) shows abnormalities in the superior and inferior quadrants.

Evaluation of the macular region is also important in glaucoma diagnosis; and OCT has become an attractive means for identifying glaucomatous macular damage. Glaucomatous damage of the macula is difficult to detect and generally overlooked or underestimated by clinicians when using the most common automated perimetry testing: 24-2 (6° grid) VF test.[98-105] Central structural damage can be missed with OCT reports based solely on circumpapillary RNFL.[78,106] Macular damage occurs early in the disease process and can take the form of arcuate defects, diffuse, widespread damage, and/or local damage, or some combination of these.[99,106-108] In eyes without other macular pathology, there is less variability and a lower likelihood of the presence of anomalous structural characteristics in the macula compared with the optic disc and peripapillary region.[47,109,110] Assessment of the macula may also avoid some limitations of circumpapillary measurements, such as interference from retinal and optic nerve head vasculature, peripapillary atrophy, and variable placement of the measurement circle around the disc.

Structures that thin and diminish in glaucoma comprise a large proportion of total macular thickness; these include RNFL and RGCs, but also inner plexiform layer.[111,112] The RGC layer is thickest in the perimacular region, and its thinning is likely responsible for the decreased total macular thickness observed in glaucomatous eyes. Preservation of central vision until late in the disease may suggest macular assessment is not useful for glaucoma detection. The macular RGC layer, however, is up to seven cells thick and contains more than 50% of the eye's RGCs,[113] and structural changes in the macula can thus easily precede detectable VF losses.[114] Additionally, changes in this layer are more likely be the result of pathologic change rather than normal variation,[115] and thus measurements of this layer could potentially be more sensitive than RNFL thickness parameters. Segmentation of the ganglion cell layer alone, however, remains very difficult due to low reflectivity.[116]

Advances in OCT have allowed for better quantitative evaluation of macular RGC damage and have enabled more detailed segmentation of the macular inner retinal layers and the entire macular thickness.[117-120] Although some studies show that macular RNFL (mRNFL) thickness in SD-OCT is less accurate than circumpapillary RNFL in glaucoma diagnosis or detecting preperimetric glaucoma, novel segmentation algorithms have increased the diagnostic utility of macular evaluation.[47,111,120-123] Parameters, such as mRNFL, ganglion cell layer with inner plexiform layer (CGIPL), and the ganglion cell complex (GCC), which includes mRNFL, ganglion cell layer, and inner plexiform layer,[117,118] can distinguish glaucomatous eyes from those of healthy subjects and can differentiate between early, moderate, and advanced glaucoma.[110,121,124-129] The GCC and GCIPL parameters from both SS-OCT and SD-OCT carry a diagnostic performance at least equal to that of circumpapillary RNFL parameters[130]; all three parameters, including mRNFL, now have comparable diagnostic performance in detection of preperimetric glaucoma.[110,124-126] Once VF losses are apparent, there is a significant association of VF defect patterns with GCIPL defect patterns.[114] The most commonly observed GCIPL and inner macular layer defect pattern in glaucoma subjects is thinning in the inferior perifoveal region, seen clinically as superior VF defects[100,111,131,132] (Figs. 2, 3).

Figure 2

Ganglion cell analysis from Cirrus-OCT (Carl-Zeiss Meditec, Dublin, CA, USA), which includes the combination of ganglion cell and inner plexiform layers (GCIPL), shows thinning in the inferior and inferotemporal perifoveal regions of both eyes.

Figure 3

Perimetric pattern deviation maps from the same patient shown in Figure 2. Defects correspond with the locations of thinning observed in ganglion cell analysis.

Macular parameters are of increasing importance in the management of glaucoma, especially given improvements in OCT technology. Recent software updates have reduced GCC segmentation errors in patients with macular degeneration, which should allow these patients to be evaluated for glaucoma with more confidence.[133,134] Posterior pole asymmetry analysis (PPAA) combines mapping of the posterior pole retinal thickness with asymmetry analysis between eyes and between hemispheres of each eye.[135] Posterior pole asymmetry analysis, although not presently having a built-in normative database, is highly reproducible and matches circumpapillary RNFL measurements in terms of diagnostic accuracy of early glaucoma.[136-,138] Spectral-domain OCT may also be superior to SS-OCT in detecting GCIPL thinning in the outer temporal zone,[119] where the glaucomatous damage commonly occurs.[100] However, SD-OCT and SS-OCT have similar glaucoma diagnosis abilities based on macular inner layer thickness analysis.[119] Ultimately, clinical assessment of the macular region is helpful in glaucoma diagnosis and evaluation and clinicians should incorporate macular scans into clinical protocols. Patients with abnormal or borderline macular structural parameters likely require close follow-up and initiation of treatment to avoid vision loss.[139] Clinicians should not rely entirely on macular parameters, however. In addition to glaucoma, other macular diseases are also common in the aging population. These conditions may affect OCT macular thickness measurements and render them useless for the evaluation of glaucoma, including diabetic retinopathy, macular edema, macular degeneration, and epiretinal membranes.

Advances in OCT technology have also allowed for higher-resolution imaging of the ONH[140] and quantification of ONH parameters, but the clinical value of OCT ONH parameters remains controversial. Compared with previous OCT technology, SD-OCT relies less on data interpolation, and this has resulted in far better delineation of ONH structures than could be achieved with TD-OCT. Studies have found that ONH parameters have an excellent ability to discriminate between normal eyes and eyes with even mild glaucoma. Parameters such as rim area, vertical rim thickness, and vertical cup to disc ratio were found to have the greatest diagnostic ability and were as good as RNFL thickness parameters in diagnosing glaucoma.[40] In some cases, ONH parameters were found to be better at initial glaucoma detection and discriminating glaucoma and glaucoma suspect subjects from normal subjects.[141,142] However, a number of studies show ONH parameters are inferior to standard circumpapillary measurements for glaucoma detection.[122,143,144] One study, although limited by using VFs as a reference standard, found that RNFL and macular parameters were significantly better for glaucoma diagnosis than ONH parameters, especially for early-stage glaucoma.[21,143] Differences in these results are at least partially a consequence of the reference standard used to select cases and controls given that a reference standard must be employed to select cases and controls for any diagnostic accuracy study.[145] There is a greater chance, for example, that patients with clearly abnormal optic disc features will be classified as cases if the reference criteria include optic disc appearance. Retinal nerve fiber layer abnormalities are not as easily detectable by clinicians.[146,147] Similarly, those with normal appearing optic discs will be declared controls. This type of reference standard introduces a bias towards favoring accuracy of topographic ONH parameters.[145] Furthermore, differences in commercially available OCT devices could at least partially explain the differences between studies, such as differences in the acquisition speed, scanning rate, spatial resolution,[142] layer detection algorithms, and analytical software. Each of the different devices, therefore, may report different RNFL thickness values and ONH measurements. Additional reasons could also contribute to variable results, including differences in the number of subjects, differences in ethnicity, or differences in the representation of the different stages of glaucoma.[40,143] Consideration of a combination of circumpapillary and ONH parameters is likely the best approach for glaucoma detection.[29,78,141]

Given recent advances in SD-OCT technology, evaluating the ONH with OCT is useful in the diagnosis and management of glaucoma. Segmentation of the ONH was greatly improved with new software, as was implementation of referencing the fovea's position and Bruch's membrane as anatomic landmarks, which allowed for better measurement of the ONH rim and RNFL thickness.[148-150] Bruch's membrane opening minimal rim width (BMO-MRW), a relatively new anatomical parameter describing the neuroretinal rim, consists of the minimum distance between the BMO and the internal limiting membrane. Bruch's membrane opening minimal rim width has a high association with glaucomatous functional changes on SAP and, compared with previous BMO methods, has a better ability to detect early glaucoma.[142,148,151] It has an advantage over other SD-OCT methods of neuroretinal thickness measurement by considering the variable orientation of rim tissue in the ONH. Rim area, however, appears to be a more useful ONH parameter for detecting early glaucoma.[142] Ultimately, assessment of circumpapillary RNFL, macular, and ONH parameters is useful for quantifying risk, diagnosis, and management of glaucoma at different levels of severity.

Detecting Progression

Detection of disease progression remains challenging in glaucoma due to the variable and slowly progressive nature of the disease, measurement variability of SAP and of imaging devices, and the lack of a commonly acceptable reference standard.[47] Some eyes show structural changes in the ONH or RNFL before any indication of glaucomatous damage can be detected with SAP.[82] Because SD-OCT is relatively new technology, only a few reports exist using SD-OCT RNFL parameters for detecting glaucoma progression.[75,152,,,155] Most progression studies use TD-OCT due to the longer follow-up period.[24,32,33,156]

Assessment of multiple OCT parameters from the macula, ONH, and RNFL is important, not only in diagnosis, but to detect disease progression and longitudinal change.[78] Retinal nerve fiber layer evaluation is less sensitive than VF when tracking progression in advanced cases due to a floor effect that occurs when the residual RGC layer has nearly diminished.[156-,158] Location of RNFL losses should also be considered when predicting VF progression.[32,156,159] Although average RNFL thickness may be the main parameter to consider when evaluating for structural progression in advanced glaucoma patients, RNFL thicknesses in the inferior quadrant and inferotemporal sector may be the most predictive of progression[33,156] (Fig. 4). Retinal nerve fiber layer thinning in the superior quadrant has also been associated with subsequent VF losses in TD-OCT.[160] Spectral-domain OCT instruments cannot be used interchangeably, however, especially for glaucoma progression assessment, due to variability in circumpapillary RNFL thickness calculations.[49] Interestingly, other studies found that average macular thickness is more sensitive than circumpapillary RNFL for detection of disease progression.[161] Additionally, analysis of the total retinal thickness (GCIPL, along with outer plexiform layer to RPE) may be more sensitive in detecting progression than circumpapillary RNFL.[162]

Figure 4

Guided progression analysis from Cirrus-OCT (Carl-Zeiss Meditec) demonstrating early glaucomatous progression. Compared with baseline examinations from December 2007, the patient has statistically significant focal RNFL loss in the inferior quadrant in the left eye. More commonly, the corresponding RNFL thickness map to “Exam 6” may show a red wedge-shaped defect in the inferior quadrant of the ONH.

Approaches that combine structure and function improve diagnosis both in cross-sectional and longitudinal investigations[163,164] and algorithms that combine structural and functional measurements will likely improve the detection of glaucoma progression.[139,163-166] Prior studies suggest that using a combination of perimetry and circumpapillary RNFL values is the best approach when monitoring for progression, especially given that SD-OCT RNFL values have a strong relationship to functional deficits.[167-172] In TD-OCT, structural progression was associated with functional progression in preperimetric, glaucoma suspect, and glaucomatous eyes[160] and eyes with significant SAP progression have higher rates of RNFL thickness loss compared with nonprogressing eyes.[173] Clinically, it can be difficult to determine whether RNFL losses that precede SAP changes reflect true progression. Important from a clinical perspective, the 24-2 VF test, although the gold standard in SAP for glaucoma evaluation, is not an ideal strategy for detecting glaucomatous damage of the macula; the 10-2 VF will often detect damage missed with the 24-2 pattern.[98,101]

As OCT evolves, it will continue to provide more accurate detection of progression and enhance our understanding of the structural pathogenesis of glaucoma, including the role of the LC in glaucoma progression. As the presumed site of axonal injury in glaucoma,[174] the LC may play a role in neuronal death seen in glaucoma. Lamina cribrosa microstructure likely provides the mechanical support to optic nerve fibers within the deep optic disc region.[175] Quantitative measurements of LC microarchitecture, such as pore diameter, pore area, and LC beam thickness, were found to have good reproducibility[56] in a multimodal SD-OCT with adaptive optics technology and offer the potential to serve as biomarkers for glaucoma progression. Using a prototype SS-OCT with 100,000 A-scan/s scanning speed and 5-μm axial resolution, Wang et al. demonstrated good reproducibility of LC parameters, including pore diameter SD, pore aspect ratio; beam thickness, pore area, beam thickness SD, and beam thickness to pore diameter ratio.[176] Lamina cribrosa microarchitecture changes have been observed with SS-OCT in glaucomatous eyes,[177] and LC pore shape and size also have been correlated with the severity and progression of glaucoma.[178,179] Additionally, the LC was found to be displaced both anteriorly and posteriorly in glaucomatous eyes compared with age-matched healthy eyes,[180-182] and thinner LC was associated with glaucoma progression.[183] Overall, the structural thinning and displacement of the LC likely cause LC pores to deform,[184] impeding axoplasmic flow within the optic nerve fibers and disrupting transport of factors crucial for the survival of RGCs.[185,186] This could lead to RGC apoptosis, contributing to glaucoma development and progression. In addition, there may be biological changes that occur due to deformations of the lamina in the axons or microglia that result in axonal stress or RGC impairment or death. Although the LC's role in glaucoma progression is yet to be fully determined, SS-OCT and SD-OCT have undoubtedly improved current understanding of the LC and its microarchitecture.[176]

Conclusions

Optical coherence tomography has changed the face of glaucoma assessment and research. Optical coherence tomography has impacted the ways that patients are diagnosed and followed clinically and remains a dynamic and evolving imaging modality. Optical coherence tomography technology and software algorithms are improving and newer technologies are continually under development,[187-189] increasing OCT's clinical utility. Clinicians should be aware of OCT's limitations and should be aware of possible scan artifacts. Clinical decisions should never be driven by OCT results alone, but should also be based on a complete ophthalmic examination and VF assessment.