Discrepancies Between Clinician and Reading Center Interpretations of OCT

The CATT study noted differences in reading center and clinician interpretations of OCT data. Why?

Dilsher Dhoot, MD • Alex Yuan, MD

Clinical trials often rely on the interpretation of data by optical coherence tomography reading centers. Differences in OCT interpretation between reading center graders and clinicians may have a large impact on the usefulness of the conclusions drawn from these clinical trials.

Macular degeneration trials are a good example. Vascular endothelial growth factor inhibitors are the first-line therapy for the treatment of exudative AMD. Although the different VEGF inhibitors that are used in AMD are approved for use at scheduled intervals, a variety of paradigms have been developed to reduce the treatment burden on patients.^1-6 Retreatment decisions in these alternative paradigms rely on clinical findings and OCT findings.

For example, patients may be treated on an “as-needed” basis if there is evidence of subretinal hemorrhage, subretinal or intraretinal fluid, or a reduction in vision. Alternately, patients may be treated using a “treat and extend” approach, whereby the interval between treatments is sequentially extended up to 10 weeks,^7,8 as allowed by the clinical and OCT findings. In the Comparison of Age Related Macular Degeneration Treatments Trial (CATT), scheduled monthly dosing of ranibizumab and bevacizumab were compared with “as-needed” dosing, with the one-year results showing that scheduled dosing was equivalent to the “as-needed” regimen for ranibizumab, although it was inconclusive for bevacizumab. The two-year data are eagerly anticipated to see whether this difference between “as-needed” bevacizumab and monthly bevacizumab persists.

One interesting finding of CATT was a discrepancy between the OCT reading centers’ interpretations of retinal fluid compared with the clinicians’ interpretations in more than 25% of cases. Of these scans, 92.8% in the ranibizumab group and 91.1% in the bevacizumab group involved the detection of fluid by reading centers in patients who were not treated by their clinicians. These results suggest that clinicians may be undertreating their patients more than 20% of the time. To understand why this difference in interpretation might occur, we will review how OCT reading center graders interpret scans and compare this with the typical clinician.

READING IS FUNDAMENTAL

The data recorded by reading centers are both objective (eg, central foveal thickness) and subjective (eg, presence of intraretinal fluid) and are predefined. Thus, graders in a reading center interpret scans based on a checklist of characteristics without any supporting clinical information and without chronological comparisons between scans.

A clinician, in contrast, will usually review the OCT along with clinical information, such as vision change and fundus examination data, and clinicians generally do not go down a checklist of items to review. This fundamental difference between reading center graders and clinicians can have a profound impact on OCT interpretation.

For example, a clinician who is examining a patient with occult choroidal neovascularization and 20/20 vision after an intravitreal injection the prior month may not look as carefully through the scans for the presence of small amounts of intraretinal fluid. A reading center grader, in contrast, will ex amine each scan without any biasing clinical information. Even a clinician participating in the study, who is given a checklist of OCT parameters to review, may be affected by this bias. There could also be ambiguous anatomical features on the OCT scan, such as outer retinal tubulation or small retinal vessels, which could be mistaken for intraretinal fluid on time-domain OCT without the benefit of serial scans for comparison.

Although the number of scans obtained is limited on a time-domain OCT, such as the Stratus (Carl Zeiss Meditec), a clinician typically does not review all of the available scans. On a macular thickness map, a single scan is often displayed and reviewed in a clinical setting. A reading center grader, in contrast, will review each radial scan obtained to look for intraretinal fluid.

On a spectral-domain OCT, such as the Cirrus (Carl Zeiss Meditec), the number of scans that are reviewed by the OCT reading center far outnumber the scans typically reviewed in a clinic. Reading center graders review all of the scans from a macular cube (typically 128 scans), whereas a clinician would likely review a single raster line scan, look at the change analysis, and perhaps review a thickness map. It would not be practical to scan through each slice of a macular cube in a clinical setting.

The one-year CATT results were based on time-domain OCT data using macular thickness maps. The second year of CATT compares spectral-domain OCT interpretations with time-domain. It will be interesting to see whether the higher-resolution scans will result in a smaller difference in interpretation or whether the increased number of scans reviewed by the graders on spectral-domain OCT will result in a growing disparity between the grader and clinician interpretations.

Figure 1 shows a comparison between the higher-resolution Cirrus scan and the corresponding Stratus scan taken from the same patient on the same day. Most graders would not recognize the presence of fluid on the Stratus scan; however, the Cirrus scan shows trace amounts of intraretinal and subretinal fluid. Figure 2 shows an example in which the typical clinical interpretation would miss the intraretinal fluid unless a more careful review of the raster or macular cube scans was performed. Reading center graders systematically review macular cube scans, whereas clinicians often do not.

Figure 1. Comparison of time-domain OCT with spectral-domain OCT. Scans taken from the same patient on the same day demonstrate the increased resolution from the spectral-domain OCT (Cirrus) compared with the time-domain OCT (Stratus). On the Stratus (A), there is a questionable area of intraretinal fluid that blends in with the background noise, whereas on the Cirrus (B), there is more definitive intraretinal and subretinal fluid. The white arrows denote the areas of fluid, and the black asterisks denote the choroidal neovascular membrane.

Figure 2. Comparison of the macular cube report, five-line raster report, and single line scans from the macular cube and raster scans: (A) The 512 x 128 macular cube report; (B) Five-line raster report; (C) Single macular cube scan; (D) Single raster scan. The clinician typically reviews only the summary reports (A and B), whereas reading center graders review each slice of the macular cube or raster scans (C and D). The arrows show areas of fluid missed on the reports.

Finally, in a clinical setting, the clinician reviews scans that have been selected by the OCT technician and either printed or sent to a computer in the examination area. A summary analysis page is often used for this purpose (Figure 2). A reading center grader reviews images using specific reading software and does not grade the images based on a single preselected scan. This difference could be another source of error.

NEED FOR CLOSER FOLLOW-UP

Despite these differences in OCT interpretation, the clinical implications are likely to be minimal when there is close follow-up. The one-year results of CATT showed there was no difference between ranibizumab scheduled dosing and “as-needed” dosing. This finding occurred despite the discrepancy observed between the clinician and reading center interpretations. If a small amount of intraretinal fluid was missed, any worsening and increase in fluid would likely be picked up the following month.

A problem could potentially exist if follow-up examinations were spaced further apart. Aflibercept is dosed every other month after the initial induction phase (the VIEW 1 and VIEW 2 studies). Intraretinal fluid missed on OCT in patients treated with aflibercept could potentially result in a much longer delay in treatment. However, because aflibercept also has a longer duration of action, it is unknown how long a delay in treatment could be tolerated before a permanent loss in vision and irreversible damage occurs.

Recent studies suggest an “as needed” regimen appears to work.^9,10 If vision loss is noted on examination or subjectively by the patient, then clinicians using spectral domain OCT might want to consider reviewing the scans directly off the reading software instead of the summary scans selected by the OCT technician.

Ultimately, the practice of medicine is an art, and the final clinical decision should not rely solely on the presence or absence of small amounts of fluid on OCT but should instead be determined by the entire clinical scenario, with the individual preferences of the patient in mind. RP

REFERENCES

1. Abraham P, Yue H, Wilson L. Randomized, double-masked, sham-controlled trial of ranibizumab for neovascular age-related macular degeneration: PIER study year 2. Am J Ophthalmol. 2010;150:315-324.e1.
2. Regillo CD, Brown DM, Abraham P, et al. Randomized, double-masked, shamcontrolled trial of ranibizumab for neovascular age-related macular degeneration: PIER Study year 1. Am J Ophthalmol. 2008;145:239-248.
3. Schmidt-Erfurth U, Eldem B, Guymer R, et al. Efficacy and safety of monthly versus quarterly ranibizumab treatment in neovascular age-related macular degeneration: the EXCITE study. Ophthalmology. 2011;118:831-839.
4. Lalwani GA, Rosenfeld PJ, Fung AE, et al. A variable-dosing regimen with intravitreal ranibizumab for neovascular age-related macular degeneration: year 2 of the PrONTO Study. Am J Ophthalmol. 2009;148:43-58.e1.
5. Boyer DS, Heier JS, Brown DM, et al. A phase IIIb study to evaluate the safety of ranibizumab in subjects with neovascular age-related macular degeneration. Ophthalmology. 2009;116:1731-1739.
6. Fung AE, Lalwani GA, Rosenfeld PJ, et al. An optical coherence tomographyguided, variable dosing regimen with intravitreal ranibizumab (Lucentis) for neovascular age-related macular degeneration. Am J Ophthalmol. 2007;143: 566-583.
7. Brown DM, Regillo CD. Anti-VEGF agents in the treatment of neovascular age-related macular degeneration: applying clinical trial results to the treatment of everyday patients. Am J Ophthalmol. 2007;144:627-637.
8. Engelbert M, Zweifel SA, Freund KB. “Treat and extend” dosing of intravitreal antivascular endothelial growth factor therapy for type 3 neovascularization/ retinal angiomatous proliferation. Retina. 2009;29:1424-1431.
9. Heier JS, Boyer D, Nguyen QD, et al. The 1-year results of CLEAR-IT 2, a phase 2 study of vascular endothelial growth factor trap-eye dosed as-needed after 12-week fixed dosing. Ophthalmology. 2011;118:1098-1106.
10. Brown DM, Heier JS, Ciulla T, et al. Primary endpoint results of a phase II study of vascular endothelial growth factor trap-eye in wet age-related macular degeneration. Ophthalmology. 2011;118:1089-1097.