Intersession repeatability of optical coherence tomography measures of retinal thickness in early age-related macular degeneration

Authors


Praveen J. Patel
Moorfields Eye Hospital
162 City Road
London EC1V 2PD
UK
Tel:  + 44 207 566 2588
Fax: + 44 207 608 6925
Email: praveenjpatel@yahoo.co.uk

Abstract.

Purpose:  To determine the intersession repeatability of Stratus optical coherence tomography (OCT) measures of retinal thickness in patients with age-related macular degeneration (AMD).

Methods:  Measurement of retinal thickness was performed over four sessions over 12 weeks using a standardized OCT protocol with the fast macular thickness map in 67 non-treated eyes of 67 patients with AMD enrolled in a clinical trial. The intrapatient standard deviation (Sw) and 95% coefficient of repeatability (CR) (inline image), expressed in μm and as a percentage of mean retinal thickness, were calculated to estimate intersession repeatability.

Results:  The CR was 32 μm for the average retinal thickness in the central 1 mm A1 subfield [95% confidence interval (CI) 31–33 μm] and 53 μm (95% CI 51–55 μm) for the centre-point thickness (CPT). When expressed as a percentage, the CR was 15% (95% CI 14–16) for the central 1 mm A1 subfield and 29% (95% CI 27–30) for the CPT measure.

Conclusion:  The average central 1 mm (A1) subfield retinal thickness measure shows good intersession repeatability in patients with stable, early AMD with poorer repeatability for the CPT measure. The results suggest that a change in Stratus OCT retinal thickness of more than 32 μm in the central A1 subfield is more indicative of true clinical change in these patients.

Introduction

Optical coherence tomography (OCT) imaging forms an important part of the structural assessment of patients with retinal disease. This non-invasive technique provides optical cross-sectional images of the retina, allowing visualization of pathology and retinal structure (Huang et al. 1991). OCT imaging is particularly important for patients with neovascular age-related macular degeneration (AMD), in whom it is used to detect quantitative and qualitative features of choroidal neovascularization (CNV) and thereby determine the need for retreatment with intravitreal therapies (Fung et al. 2007). In addition, changes in OCT-determined retinal thickness measurements are used as an end-point in clinical trials and it is therefore important to determine the repeatability of retinal thickness measurements to distinguish true clinical change from measurement error (Csaky et al. 2008).

Previous reports of the repeatability of OCT-derived retinal thickness measurements show greater variability in patients with retinal disease (Krzystolik et al. 2007; Danis et al. 2008) than in normal patients (Paunescu et al. 2004; Polito et al. 2005). The repeatability of OCT-derived thickness measures in one type of macular disease may differ from that of another: previous work from our group reported greater variability for intrasession measurements in neovascular AMD patients (Patel et al. 2008b) in comparison to studies in patients with diabetic macular oedema (Krzystolik et al. 2007). Also, the repeatability of measurements taken at the same visit (intrasession repeatability) may be different from those taken at different visits (intersession repeatability). Given the increasing use of retinal thickness measurements in patients with AMD, it is important to report repeatability measurements in this group of patients.

We have previously reported the intrasession repeatability of OCT-derived retinal thickness measurements in patients with neovascular AMD (Patel et al. 2008b). However, it is not possible to investigate intersession repeatability directly in this group of patients because of clinical or subclinical disease progression between visits. The aim of this study was to investigate the intersession repeatability of Stratus OCT-derived retinal thickness measures in a cohort of patients with stable, early AMD. The results from this study will provide valuable information regarding the variability of these measurements between visits in this elderly group of patients. In addition, the values obtained may be useful when distinguishing clinical change (onset of late AMD – geographic atrophy or CNV) from measurement variability in this group of patients. Furthermore, it may be reasonable to apply the estimates obtained when assessing disease recurrence in patients with treated, quiescent neovascular AMD with good visual acuity and well-preserved retinal architecture.

Materials and Methods

Patients

Data from the untreated eye of patients enrolled in the Avastin® (bevacizumab) for Choroidal Neovascularization Trial (Patel et al. 2008a) were used in this repeatability study. The ongoing clinical trial is a prospective, randomized controlled trial investigating the safety and efficacy of intravitreal bevacizumab (Avastin; Genentech Inc., South San Francisco, California, USA) in the treatment of neovascular AMD. All patients had consented to OCT imaging and the research followed the tenets of the Declaration of Helsinki. In addition, approval for this research had been obtained from both the Trial Steering Committee and the Research Governance Committee of Moorfields Eye Hospital. OCT-determined central 1 mm (A1) subfield retinal thickness and centre-point thickness (CPT) measurements at baseline, week 1, week 6 and week 12 visits from a total of 124 eyes of 124 patients were available for analysis. We identified eyes with stable, non-neovascular, early AMD from this cohort and included these in the analysis. For the purposes of this study, we defined early AMD as eyes with at least one hard druse in the macula without evidence of CNV, pigment epithelial detachment or geographic atrophy on clinical examination, OCT and fluorescein angiography. Clinical examination, fluorescein angiography and OCT imaging were used to detect and exclude eyes with evidence of disease progression over the four patient visits because the assessment of repeatability of measurements taken over 12 weeks assumes no clinical or subclinical change in disease status over this period. In addition, because the eyes included in this repeatability study had non-neovascular, early AMD, the retinal thickness would not be affected by therapy delivered to the contralateral eye. For analysis purposes, eyes were classified into two AMD subgroups: eyes with small to intermediate (< 125 μm) drusen only and eyes with large drusen (≥ 125 μm with or without pigment changes). The best-corrected visual acuity was also measured at each visit by clinical trial accredited optometrists using Early Treatment of Diabetic Retinopathy Study (ETDRS) charts.

Equipment

All OCT imaging was performed using the commercially available Stratus OCT machine (Carl Zeiss Meditec Inc., Dublin, California, USA) with software version 4.0. This is the third generation of OCT machine and provides an axial resolution of < 10 μm. The OCT machine was serviced regularly in line with manufacturer recommendations by authorized technicians and personnel from Carl Zeiss Meditec Inc. to ensure that the machine was calibrated and operating correctly.

Scanning

All scanning was performed by a single technician certified by image reading centres for OCT scanning in pharmaceutical-company-sponsored AMD clinical trials. In line with other studies, the fast macular thickness map (FMTM) protocol was used to assess retinal thickness. The FMTM protocol uses six high-speed 6-mm radial lines (oriented 60° apart) to delineate macular anatomy and pathology. This scan mode enables all six line scans to be acquired in a continuous, automated sequence within 1.92 seconds, with each of the six lines composed of 128 equally spaced transverse sampled locations (total of 128 × 6 lines or 768 sampled points). At each of these locations, the signal is sampled axially at 1024 equal intervals over a depth of 2 mm. Because of its short acquisition time, the FMTM protocol is believed to be less prone to errors caused by unstable fixation – an important consideration in the assessment of patients with AMD. However, this rapid acquisition of scans is at the expense of scan resolution in the transverse plane. All OCT imaging was performed using a standardized protocol. The patient was aligned correctly with the OCT machine and was asked to look at an internal fixation light. If no light was seen, the patient was asked to look straight ahead by use of an external fixation light to ensure that the scans were taken through the fovea. The technician was experienced in identifying common artefacts in OCT images and scan sets were reacquired as needed to optimize scan quality. Poor-quality scans with artefact or a signal strength < 7 were discarded and the technician was instructed to save only high-quality, well-centred scans with signal strength > 6 (although such standards were not always achieved). Once image quality had been optimized, the technician acquired several FMTM scans with data from the best OCT scan set saved and used for analysis. Each scan was analysed using the onboard Stratus OCT software (version 4.0) with automated segmentation of retinal boundaries and quantitative measurement of retinal thickness. The macular thickness map analysis uses data from all six linear scans to interpolate the retinal thickness measurements across circular ETDRS sectors with nine sectoral thickness values for circles with diameters of 1, 3 and 6 mm (Fig. 1). The automated measurements (in mm2) of retinal thickness in the A1 subfield and the automated CPT were noted for each patient and each visit. The A1 subfield measure represents an average retinal thickness for the central 1 mm circular subfield while the CPT is a measure of retinal thickness at a single point at the centre of the subfield (at the intersection of all six line scans). Patients with retinal maps with a low analysis confidence message on more than one visit (suggesting failure of the segmentation software) were excluded from the repeatability analysis, as were patients with disease progression (or development of CNV).

Figure 1.

 Macular subfields for Stratus optical coherence tomography fast macular thickness map protocol analysis. The central A1 field has a diameter of 1 mm, fields A2–A5 are zones of a circle 3 mm in diameter and fields A6–A9 are zones of a circle 6 mm in diameter.

Statistical analyses

Statistical analyses were performed using spss (version 12.0.1; SPSS Inc., Chicago, Illinois, USA). The mean retinal thickness for the entire cohort of patients at each visit was calculated for the A1 subfield and the CPT. The difference between retinal thicknesses across the four visits was analysed using the non-parametric Friedman test. In line with recommendations outlined by Bland & Altman (1996), the standard deviation (SD) of retinal thickness measurements for individual patients was calculated and plotted against the mean value for each patient (for all four measurements). The intrapatient SD (Sw), derived from the residual mean square, was used to calculate the 95% coefficient of repeatability (CR) defined by Bland and Altman as inline image or 2.77Sw. The difference between two intersession measurements for the same patient is expected to be less than the CR for 95% of pairs of observations. In addition, 95% confidence limits for the CR were calculated. The width of the 95% CI for the population within-patient SD is inline image, where n is the sample size and m is the number of observations for each patient (Bland 2004).

For consistency with other studies, the CR was also expressed as a percentage of mean retinal thickness. This was carried out for the entire cohort of patients with additional values calculated for the two AMD diagnostic categories.

Results

Sixty-seven eyes of 67 patients were identified with stable early AMD from the cohort of 124 patients; a total of 268 OCT-derived retinal thickness measurements taken over four visits were included in the analysis. Twenty-eight male patients and 39 female patients provided 24 left eyes and 43 right eyes. The number of eyes in each AMD category and the mean age and visual acuity of patients is shown in Table 1. The mean time (± SD) in days between the baseline and week 1 OCT measurements was 7 (± 3) days. The mean interval between baseline and week 12 visits was 85 (± 8) days.

Table 1.   Diagnoses, age and visual acuity of patients.
DiagnosisNumber of eyesMean age in years (SD)Mean visual acuity in ETDRS letters (SD)
  1. SD, standard deviation; ETDRS, Early Treatment of Diabetic Retinopathy Study; AMD, age-related macular degeneration.

Early AMDSmall/intermediate drusen3277 (7)82 (5)
Early AMDLarge drusen with or without pigment change3579 (7)77 (8)
Total 6778 (7)79 (8)

The mean retinal thickness of patients at each visit is shown in Tables 2 and 3 for each measure of central retinal thickness, with no significant difference between visits for either the A1 subfield thickness ( p = 0.07) or CPT (p = 0.11). The plot of intrapatient SD against mean retinal thickness for the A1 subfield is shown in Fig. 2, with the plot of intrapatient SD against mean retinal thickness for the CPT measurement shown in Fig. 3. Both plots show that, as expected, this cohort of early AMD patients had a relatively narrow range of retinal thicknesses in the 150–250 μm range. There was no relationship between intrapatient SD and retinal thickness. The CR for the average retinal thickness in the A1 subfield (Table 4) shows similar repeatability for patients with small to intermediate drusen and for patients with large drusen (with or without pigment change). The CR for the CPT measure is shown in Table 5. The overall CR for this cohort of stable, early AMD patients for all visits was 32 μm for the A1 subfield (95% CI 31–33 μm) and 53 μm (95% CI 51–55 μm) for the CPT. When expressed as a percentage, the CR was 15% (95% CI 14–16) for the A1 subfield and 29% (95% CI 27–30) for the CPT measure.

Table 2.   Median retinal thickness (μm) for central 1 mm (A1) subfield at each visit.
DiagnosisAverage retinal thickness in A1 subfield, median (interquartile range) (μm)p-value (Friedman test)
Visit week
01612
  1. AMD, age-related macular degeneration.

Early AMDSmall/intermediate drusen (n = 32)213 (22)212 (23)211 (22)213 (20)0.36
Early AMDLarge drusen with or without pigment change (n = 35)214 (26)215 (24)211 (22)210 (20)0.22
Total (n = 67)214 (25)213 (23)211 (22)211 (20)0.07
Table 3.   Median retinal thickness (μm) for centre-point thickness (CPT) at each visit.
DiagnosisMedian CPT retinal thickness (interquartile range) (μm)p-value (Friedman test)
Visit week
01612
  1. AMD, age-related macular degeneration.

Early AMDSmall/intermediate drusen (n = 33)187 (27)186 (26)184 (30)189 (30)0.54
Early AMDLarge drusen with or without pigment change (n = 34)189 (28)190 (30)184 (29)185 (25)0.18
Total (n = 67)188 (28)188 (28)184 (29)187 (27)0.11
Figure 2.

 Plot of individual patients’ standard deviation against mean central 1 mm retinal thickness.

Figure 3.

 Plot of individual patients’ standard deviation against mean centre-point retinal thickness.

Table 4.   Coefficient of repeatability (CR) for the central 1 mm subfield measurement.
DiagnosisNumber of eyesCoefficient of repeatability 
In μmWhen expressed as percentage of mean retinal thickness
  1. AMD, age-related macular degeneration.

Early AMDSmall/intermediate drusen323215
Early AMDLarge drusen with or without pigment change353316
Total673215
Table 5.   Coefficient of repeatability (CR) for the centre-point thickness (CPT) measurement.
DiagnosisNumber of eyesCoefficient of repeatability 
In μmExpressed as percentage of mean retinal thickness
  1. AMD, age-related macular degeneration.

Early AMDSmall/intermediate drusen325630
Early AMDLarge drusen with or without pigment change355127
Total675329

Discussion

The introduction of OCT-based retinal imaging into clinical practice has led to significant advances in the assessment of patients with vitreoretinal disease. The introduction of new therapies for neovascular AMD, which were initially developed and used for patients with macular oedema, has seen OCT become an essential tool in the evaluation of patients with AMD. The assessment of disease activity, disease progression and the response to treatment in patients with AMD involves a complex analysis of both qualitative features of retinal abnormalities and OCT-determined retinal thickness change. Indeed, change in retinal thickness is used as one of the triggers for retreatment with ranibizumab (Fung et al. 2007) and as an end-point in clinical trials (Csaky et al. 2008). Therefore, it is important to establish the repeatability of OCT-derived retinal thickness measurements taken at different visits (intersession repeatability) in order to distinguish clinical change from measurement variability with longitudinal follow-up of patients. This is especially important in patients with AMD (for both neovascular and non-neovascular disease types) in view of the extensive research showing new strategies for treatment and the need to detect and measure treatment response using patient follow-up.

Several studies have investigated the repeatability of Stratus OCT-determined retinal thickness in normal patients and in patients with retinal disease (Table 6). These suggest that repeatability may vary across disease states and that the repeatability of measurements taken at a single visit (intrasession repeatability) may differ from the repeatability of measurements taken at different visits (intersession repeatability). The factors underlying this are unclear but may relate to differences in visual acuity, fixation stability and the age of the patient. Studies assessing the intersession repeatability of macular thickness have reported Bland–Altman 95% CR ranging from 10 to 35 μm for healthy individuals (Paunescu et al. 2004; Polito et al. 2005; Eriksson & Alm 2008; Eriksson et al. 2008) and from 23 to 38 μm for patients with diabetic macular oedema (Polito et al. 2005; Krzystolik et al. 2007). However, there have been few reports on the repeatability of OCT-derived retinal thickness measurements in patients with AMD. A previous study by our group reported an intrasession 95% CR for the A1 subfield of 67 μm for patients with neovascular AMD (Patel et al. 2008b). It is not possible to investigate the intersession repeatability of retinal thickness measures in patients with active neovascular AMD because of the effects of disease progression between visits and the need for prompt treatment. The aim of this study was to report the intersession repeatability of these OCT retinal thickness measurements in patients with stable, early AMD. We report a 95% CR of 32 μm for the A1 subfield, with similar repeatability for the two subgroups of early AMD patients. As in other studies (Patel et al. 2008b), we found CPT to be a poorly repeatable measure despite the relatively good visual acuity in this cohort of patients and would not recommend using it when assessing retinal thickness change between visits.

Table 6.   Coefficients of repeatability (CR) for Stratus optical coherence tomography-derived macular thickness measurements from previous studies.
StudySample size and diagnosisIntrasession or intersession repeatability? CR for central A1 subfield (μm or % of mean thickness)
  1. AMD, age-related macular degeneration; CSMO, clinically significant macular oedema; MO, macular oedema; SD, standard deviation.

Polito et al. 200510 eyes of 10 normal patients
15 eyes of 15 diabetic patients with CSMO
Intrasession and intersession3–4% (9–10 μm) intrasession, 5.5% (12 μm) intersession for normal patients;
6–7% (23–27 μm) intrasession for CSMO patients
Paunescu et al. 200410 eyes of 10 normal patientsIntersession4.9 μm intrasession SD (CR 10 μm) to 8.5 μm intersession SD (CR 17 μm)
Krzystolik et al. 20071204 scan pairs from 212 eyes of 107 patients with CSMOIntrasession38 μm
Browning & Fraser 200422 eyes of 19 patients (normals and stable macular disease)Intersession35–37 μm
Danis et al. 2008281 scan pairs (139 normal, 66 AMD, 76 MO)Intrasession27 μm (10 μm for 87 patients with thickness ≤ 175 μm and 38 μm for 48 patients with thickness ≥ 426 μm)
Eriksson et al. 200854 eyes of 54 children, 3 scans per eyeIntrasession10.2 μm (calculated from published coefficient of variation)
Eriksson & Alm 200845 eyes of 45 patientsIntrasession12.4 μm
Patel et al. 2008b50 scan pairs from 50 eyes of 50 patients with neovascular AMDIntrasessionSD 67 μm (50 μm after excluding 9 patients with segmentation failure)

The clinical utility of this data may be when assessing patients with early AMD for early signs of progression to late AMD (geographic atrophy or CNV). The results suggest that a retinal thickness change of > 32 μm in the A1 subfield is indicative of true clinical change rather than measurement variability. Moreover, in clinical practice when assessing eyes with neovascular AMD undergoing therapy, we evaluate changes in macular oedema and retinal structure in eyes that for the most part have relatively well-preserved retinal and subretinal architecture. These eyes may have a central macular thickness of < 250 μm with relatively normal foveal structure and good vision. Because it is not possible to measure the intersession repeatability of retinal thickness measurements in eyes with neovascular AMD (as a result of the possibility of clinical or subclinical disease progression between measurement visits), it may be reasonable to use intersession repeatability values obtained from patients with stable, early AMD when assessing OCT-determined retinal thickness change in patients with treated, quiescent neovascular AMD (nAMD) and an A1 subfield retinal thickness of < 250 μm with good visual acuity undergoing OCT imaging to detect disease recurrence. Indeed, if we apply our repeatability estimate to those patients with treated, quiescent nAMD with relatively normal retinal architecture, our results suggest that we may use a change criterion of > 32 μm (15% of retinal thickness) as part of the assessment for disease recurrence. However, our results would not be applicable to eyes with more advanced neovascular AMD disease with retinal thickness > 250 μm, poor visual acuity and more extensive alterations of the outer highly reflective band (irregularity from fibrosis or thick neovascular tissue and retinal pigment epithelial detachment).
The factors that underlie the variability of these measurements include both patient- and disease-related factors and instrument- or measurement-related factors. Patients with AMD may have poor or variable fixation, making it difficult to ensure that exactly the same retinal point is imaged at successive visits. This problem of unstable fixation will have the greatest effect in reducing repeatability in patients with marked topographical variation in retinal thickness across the area scanned. In patients with pathology resulting in more uniform thickening of the macula (minimal topographic variation), unstable fixation will have less effect on the repeatability of measurements because neighbouring retinal points are more likely to have similar or identical retinal thicknesses.

Measurement-related factors limiting the repeatability of retinal thickness measurements include variability in segmentation of inner and outer retinal boundaries between visits and changes in image quality, which may all limit repeatability. Faster spectral-domain OCT imaging technology (Wojtkowski et al. 2004) and eye-tracking capabilities (Hammer et al. 2005) may help to improve the repeatability of retinal thickness and volume in this important patient group by ensuring point-to-point registration between visits.

The strengths of this study include the large sample size and the number of measurements taken for each patient. The use of scans taken by a single OCT technician accredited for clinical trial work to eliminate interobserver variability is another strength. One limitation of the study is that only the CPT and the A1 subfield data were available for analysis. The fact that this study was not designed specifically as a repeatability study but utilized data from an ongoing clinical trial is a further limitation. However, this may also be viewed as a strength: although studies designed specifically to measure repeatability may achieve an extremely high degree of repeatability using researchers dedicated to this goal, the values may not be applied easily to other settings. The technician in this study was carrying out imaging adhering to a standardized clinical trial protocol but without the knowledge that the repeatability of measurements would be assessed formally. This better reflects a true clinical trial setting, in which many measurements are taken by observers following set protocols but often without formal assessment of intersession repeatability.

In summary, this is the first study to report the intersession repeatability of OCT-derived retinal thickness measurements in a large cohort of patients with early, stable AMD. Our results are useful when attempting to distinguish clinical change from measurement error in patients with AMD. In addition, it may be reasonable to apply these results as part of a clinical algorithm to detect disease recurrence in eyes with quiescent CNV, good visual acuity and well-preserved retinal architecture.

Acknowledgements

Financial support came from the Special Trustees of Moorfields Eye Hospital.

Ancillary