September 2012
Volume 53, Issue 10
Free
Retina  |   September 2012
Characterization of Vitreoretinal Interface Disorders Using OCT in the Interventional Phase 3 Trials of Ocriplasmin
Author Affiliations & Notes
  • Francis Char DeCroos
    From the Wills Eye Institute/Mid Atlantic Retina, Philadelphia, Pennsylvania; the
    Duke University Eye Center, Durham, North Carolina; and
  • Cynthia A. Toth
    Duke University Eye Center, Durham, North Carolina; and
  • Francisco A. Folgar
    Duke University Eye Center, Durham, North Carolina; and
  • Stephen Pakola
    ThromboGenics NV, Leuven, Belgium.
  • Sandra S. Stinnett
    Duke University Eye Center, Durham, North Carolina; and
  • Cynthia S. Heydary
    Duke University Eye Center, Durham, North Carolina; and
  • Russell Burns
    Duke University Eye Center, Durham, North Carolina; and
  • Glenn J. Jaffe
    Duke University Eye Center, Durham, North Carolina; and
  • Corresponding author: Glenn J. Jaffe, Duke Eye Center, Duke Reading Center, DUMC Box 3802, Durham, NC 27710; jaffe001@mc.duke.edu
Investigative Ophthalmology & Visual Science September 2012, Vol.53, 6504-6511. doi:10.1167/iovs.12-10370
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Francis Char DeCroos, Cynthia A. Toth, Francisco A. Folgar, Stephen Pakola, Sandra S. Stinnett, Cynthia S. Heydary, Russell Burns, Glenn J. Jaffe; Characterization of Vitreoretinal Interface Disorders Using OCT in the Interventional Phase 3 Trials of Ocriplasmin. Invest. Ophthalmol. Vis. Sci. 2012;53(10):6504-6511. doi: 10.1167/iovs.12-10370.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose.: We determined the reproducibility of a novel optical coherence tomography (OCT) protocol designed to evaluate formally vitreoretinal interface abnormalities on scans obtained during two phase 3 studies of intravitreal ocriplasmin to treat symptomatic vitreomacular adhesion with or without macular hole.

Methods.: Certified technicians obtained time-domain OCT scans that included a macular thickness map (MTM), Fast MTM, and three high resolution linear scans: one 10 mm horizontal and one 10 mm vertical through the optic nerve head (ONH), and one 10 mm 5-degree-offset through the ONH and fovea. Reading Center teams graded all 3695 scans from 652 study eyes for pre-established quantitative and morphologic features. Grading reproducibility at baseline and follow-up visits was tested for presence of vitreomacular adhesion (VMA), width of vitreous adhesion (focal <1500 μm versus broad >1500 μm), presence and minimum width of full thickness macular hole (FTMH), and presence of epiretinal membrane (ERM).

Results.: Team grading reproducibility for VMA (kappa 0.91, 95% confidence interval [CI] 0.81–1.00), broad versus focal width of vitreous adhesion (kappa 0.87, 95% CI 0.78–0.95), FTMH (kappa 0.87, 95% CI 0.78–0.95), and ERM (kappa 0.87, 95% CI 0.78–0.95) was high. Percent agreement was 97%, 92%, 95%, and 82% for VMA, vitreous adhesion width, FTMH, and ERM, respectively. For repeated measurements of FTMH width, the intraclass correlation was 0.89 (95% CI 0.85–0.93), and the mean paired difference between grading team measurements was 34.4 μm (95% limits of agreement −149.5–218.2 μm).

Conclusions.: Quantitative and morphologic vitreoretinal interface features were assessed reproducibly using a newly developed OCT scan acquisition and grading protocol. This protocol will be useful to evaluate OCT endpoints in future clinical trials, and can facilitate identification of vitreoretinal interface pathology during care of individual patients. (ClinicalTrials.gov number, NCT00781859 and NCT00798317.)

Introduction
Vitreous adhesion to the retinal internal limiting membrane (ILM) is mediated in part by laminin and fibronectin. 1,2 Ocriplasmin (Microplasmin; ThromboGenics, Leuven, Belgium) is a recombinant truncated human plasmin with intact protease activity. When given as an intravitreal injection, ocriplasmin degrades fibronectin and laminin, 3 and is a promising method to induce posterior vitreous detachment (PVD). 47  
The Microplasmin IntraVitreal Injection for Traction Release without Surgical Treatment (MIVI-TRUST) program investigated the safety and efficacy of ocriplasmin to treat symptomatic vitreomacular adhesion with or without macular holes. To be enrolled, patients were required to have vitreomacular adhesion (VMA), defined as focal adherence of the posterior hyaloid to the macula with detached vitreous on both sides of the adhesion on at least one optical coherence tomography (OCT) image. VMA was considered symptomatic if, in the opinion of the trial investigator, it caused decreased visual function such as reduced visual acuity, metamorphopsia, or other visual complaint. Nonsurgical resolution of VMA on OCT was the primary ocriplasmin study endpoint, and closure without vitrectomy of preexisting macular hole, as determined by OCT was a secondary endpoint. The Duke Reading Center independently determined both endpoints through masked grading of images generated during the study. 
OCT was used to determine primary and secondary study endpoints. This imaging modality was suited ideally to evaluate vitreomacular adhesion in this trial, as it displays spatial relationships between the posterior vitreous and inner retina in a manner that otherwise is unachievable with a noninvasive technique. Likewise, OCT has proven valuable to demonstrate pathologic conditions associated with incomplete PVD, such as epiretinal membrane (ERM), 8,9 macular hole, 10,11 and tractional macular edema. 8,12 Incomplete separation of the posterior vitreous from the macula also may exacerbate diseases, such as diabetic macular edema. 13,14 retinal schisis in high myopes, 15 and neovascular age-related macular degeneration. 1618  
While OCT undoubtedly is helpful to characterize vitreoretinal interface disorders, no consensus exists for a standardized vitreoretinal interface classification scheme based on OCT. In a study of 209 healthy eyes, Uchino et al. proposed a 5-stage scale that ranged from “no PVD” to “complete PVD.” 19 This protocol was modified by Johnson and applied to 43 eyes with various vitreoretinal interface disorders. 20 More recently, one group has proposed a 3-category scale to correlate vitreous morphology with post-vitrectomy outcomes, 12 while others have described a system that includes width of vitreomacular adhesion and presence of morphologic findings on OCT. 8 From the varied classification methods described in the literature, it is clear that a standardized imaging protocol to depict and analyze systematically vitreoretinal interface disorders is needed, and would be useful especially when OCT is used to evaluate treatment effects in interventional clinical trials for vitreoretinal interface disorders. 
Since reproducibility of prior protocols for OCT-based analysis of vitreoretinal disorders is not well known, it would be important to determine the precision of any proposed method to acquire and analyze OCT scans in eyes with vitreoretinal interface pathology. This analytical precision is of particular importance to the MIVI-TRUST program, since baseline VMA was an eligibility requirement for entry into the ocriplasmin phase 3 trials and resolution of VMA as determined by OCT was a primary study endpoint. OCT also may yield important prognostic information as ocriplasmin was more likely to resolve VMA and macular hole in eyes without ERM and in those with focal vitreous adhesion, again emphasizing the need for reliable OCT grading. 4,21 We describe a method to acquire and evaluate systematically OCT images of eyes with vitreoretinal interface disorders, and also determine the reproducibility of this OCT grading protocol as used during two pivotal phase 3 interventional trials. 
Methods
Trial Overview
The MIVI-TRUST program consists of two pivotal multicenter, randomized, phase 3 trials that investigated the safety and efficacy of ocriplasmin compared to placebo injection for the treatment of symptomatic vitreomacular adhesion with or without macular hole (in the public domain, ClinicalTrials.gov identifier: NCT00781859 and NCT00798317). For this study, VMA was denoted when the posterior hyaloid adhered to the macula with vitreous separation on both sides of the adhesion on at least one OCT image. The MIVI-TRUST program included eyes with VMA resulting in decreased visual function and visual acuity worse than or equal to 20/25. For the 6-month follow-up duration, post-injection eyes were followed with serial OCT scans at each study visit. Additionally, study-site echographers performed B-scan ultrasonography during the MIVI-TRUST program, however these images were not evaluated by the Reading Center. All experimental procedures adhered to the tenets of the Declaration of Helsinki, appropriate Institutional Review Board (IRB) approval was obtained, and all participants engaged in an informed consent process and signed a written consent document before enrollment in the MIVI-TRUST program. 
OCT Acquisition
OCT scans were acquired in a standardized fashion as follows. All scans were acquired in dilated eyes by certified technicians with Stratus OCT machines using software version 4.0 or greater. To obtain certification, technicians were required to submit 32 OCT scans depicting partial or complete vitreous separation for feedback by the Reading Center regarding scan quality and avoidable artifacts. To obtain optimal visualization of the vitreoretinal interface and retina, the scan protocol also emphasized appropriate OCT scan focus, saturation, and line placement. For the duration of the study, an automated system relayed Reading Center feedback to all OCT technicians for each graded image through reports that detailed scan placement, quality, and individually identified submissions of concern. 
At each study visit, technicians performed 5 distinct scan protocols on each study eye. All eyes were imaged using the Stratus fast macular thickness map (FMTM) and macular thickness map (MTM) protocols. These protocols each included 6 radial lines of 6 mm length placed at the foveal center at 30-degree rotational increments. Three custom, high-resolution (512 A-scans per tomographic line) line scans that each were 10 mm in length also were obtained on all eyes; two that produced a “cross-hair” pattern were performed through the optic nerve, one vertically at 90 degrees and the other horizontally at 180 degrees. The third “offset” scan through the optic nerve center and foveal center was angled 5 degrees inferiorly from the optic nerve, yielding a 5-degree angle for the right eye and a 355-degree angle for the left eye (Fig. 1). Scans were de-identified and labeled with a unique code, and then submitted to the Reading Center. Scans were submitted from the 7 study visits and included OCT images recorded at baseline; injection day; post-injection days 7, 14, and 28, and post-injection months 3 and 6. 
Figure 1. 
 
Standardized OCT imaging protocols used to analyze the vitreoretinal interface during the MIVI 3 trial. White dashes: 6 radial line scans 6 mm in length were acquired using the FMTM and MTM protocols. White solid line: custom 10 mm horizontal and vertical line scans performed through the center of the ONH. Black dashes: custom 10 mm line scan through both the ONH and fovea angled at 5 degrees. All scans are high resolution (512 A-scans per tomographic line) except for those from the FMTM protocol (128 A-scans per tomographic line).
Figure 1. 
 
Standardized OCT imaging protocols used to analyze the vitreoretinal interface during the MIVI 3 trial. White dashes: 6 radial line scans 6 mm in length were acquired using the FMTM and MTM protocols. White solid line: custom 10 mm horizontal and vertical line scans performed through the center of the ONH. Black dashes: custom 10 mm line scan through both the ONH and fovea angled at 5 degrees. All scans are high resolution (512 A-scans per tomographic line) except for those from the FMTM protocol (128 A-scans per tomographic line).
OCT Grading
The 3695 Stratus OCT scans from 652 study eyes were graded by certified readers at the Duke Reading Center. For this study, the fovea on any individual OCT line scan was defined as the horizontal region 1 mm in width with midpoint at the foveal center. First, quality of each OCT scan was determined. Scan quality was termed “acceptable” if the scan was available, positioned correctly, and adequately saturated. The designation “not interpretable” was applied if one or more line images were missing, positioned incorrectly, or very poorly saturated and prevented grading a feature as present or absent. 
Our protocol sought to identify reproducibly vitreoretinal interface findings of potential clinical relevance. To this end each OCT scan was analyzed for presence of full thickness macular hole (FTMH) and ERM (Fig. 2). Each of 15 line scans (6 from the FMTM protocol, 6 from the MTM protocol, and 3 from custom 10 mm protocols) was evaluated, and then a grade was assigned for each predetermined morphologic feature. For each morphologic feature analyzed, one of the following terms was assigned: feature present, feature absent, or not interpretable (due to poor saturation, incorrect placement, or scan absence). Standardized OCT images were chosen to reference all morphologic characteristics of interest. 
Figure 2
 
Representative morphologic features noted on OCT images derived from the macular thickness map MTM protocol. (a) FTMH. (b) ERM.
Figure 2
 
Representative morphologic features noted on OCT images derived from the macular thickness map MTM protocol. (a) FTMH. (b) ERM.
Morphologic feature characteristics were subcategorized further during grading. Retinal deformation at any vitreous attachment sites or ERM and whether the central 1 mm (horizontal dimension) of the retina was deformed were recorded. In addition, a notation was made to indicate whether ERM was located at any point of vitreous adhesion. If a macular hole was present, the presence of intraretinal fluid at the macular hole lateral borders and/or vitreous adherent to the macular hole lateral borders also was recorded. 
FTMH was characterized by two measurements performed with software-based calipers on images derived from the MTM protocol. All 6 radial images were reviewed, and the maximum FTMH width at the level of the RPE was recorded (Fig. 3a). Similarly, the minimum FTMH width at any location along the hole was determined on all 6 radial images. The largest of these minimum widths was recorded (Fig. 3b). Eyes with minimum FTMH width measurement greater than 400 μm at the baseline visit were not eligible for this trial. 
Figure 3. 
 
Representative OCTs demonstrating caliper placement for each software-based measurements of FTMH width. (a) Maximum FTMH width at the level of the RPE. (b) Minimum FTMH width.
Figure 3. 
 
Representative OCTs demonstrating caliper placement for each software-based measurements of FTMH width. (a) Maximum FTMH width at the level of the RPE. (b) Minimum FTMH width.
Our grading system categorized the interaction between the posterior hyaloid and inner retina. VMA was denoted when the posterior hyaloid adhered to the macula with vitreous separation on both sides of the adhesion on at least one of the radial line scans. Subtypes of VMA included vitreous attachment at the fovea and optic nerve head (ONH) with intervening separation (Fig. 4a), or vitreous attachment only at the fovea (Fig. 4b). Posterior vitreous interaction with the retina could be classified only after reviewing all scans from a study visit. A reader could classify scans from a study eye as depicting VMA only after observing at least one image depicting posterior vitreous attachment to the macula with separation on both sides of the attachment (Figs. 4a, 4b). The vitreoretinal interface configuration could not be termed VMA based only on a single image depicting no visible vitreous separation between the fovea and ONH, but visible vitreous separation outside this zone (Fig. 4c), or an image depicting vitreous attachment between the fovea and ONH with intervening separation, but with no vitreous separation visible temporal to the fovea (Fig. 4d). 
Figure 4
 
OCT images depicting the types of vitreomacular adhesion observed during the MIVI 3 trial. (a, b) Examples of VMA. (a) Vitreous attachment at the ONH, and fovea and temporal macula with intervening separations. (b) Vitreous attachment at the fovea with separation on each side. (c) No visible vitreous from the ONH to the fovea with separation outside this zone (not considered VMA in study). (d) Vitreous attachment at the foveal edge of small FTMH and ONH with intervening separation and no visible vitreous separation temporal to the fovea (not considered VMA in study).
Figure 4
 
OCT images depicting the types of vitreomacular adhesion observed during the MIVI 3 trial. (a, b) Examples of VMA. (a) Vitreous attachment at the ONH, and fovea and temporal macula with intervening separations. (b) Vitreous attachment at the fovea with separation on each side. (c) No visible vitreous from the ONH to the fovea with separation outside this zone (not considered VMA in study). (d) Vitreous attachment at the foveal edge of small FTMH and ONH with intervening separation and no visible vitreous separation temporal to the fovea (not considered VMA in study).
Several vitreous states other than VMA were observed during the trial. Upon review of the complete set of scans, a lack of visible vitreous adhesion to the macula or optic nerve was categorized as no visible vitreous separation or adhesion (Figs. 5a, 5b). In the MIVI-TRUST trials, eyes initially had vitreomacular adhesion, with incomplete vitreous separation. Accordingly, in subsequent visits, if OCT scans appeared as in Figure 5b (no visible vitreous attachment or separation), then by definition the vitreous had separated and retracted out of the field of view. Another vitreous state observed during the course of the trial was vitreous attached at the ONH only (Fig. 5c). Finally, if the scan position, lack of availability, or poor saturation precluded evaluation of vitreomacular adhesion, then “not interpretable” was recorded. 
Figure 5
 
OCT images depicting vitreous morphologies other than vitreomacular adhesion observed during the MIVI 3 trial. (a, b) No visible vitreous attachment or separation. (c) Vitreous attached at the ONH only.
Figure 5
 
OCT images depicting vitreous morphologies other than vitreomacular adhesion observed during the MIVI 3 trial. (a, b) No visible vitreous attachment or separation. (c) Vitreous attached at the ONH only.
For all OCT scans that had VMA, the width (horizontal dimension) of the vitreous adhesion was measured for each line scan using software-based calipers. For multifocal attachments with intervening separations, each of the individual attachment widths was summed to arrive at the final total attachment width. Vitreous adhesion width greater than 1500 μm on one or more image was termed “broad,” while vitreous adhesion width less than or equal to 1500 μm on all images was denoted “focal.” 8 Vitreous adhesion width was measured preferentially from images produced by the MTM protocol, but available FMTM images could be substituted when MTM images were not interpretable. 
Team-Based OCT Evaluation
Two masked readers first graded all scans independently. Next, a separate data specialist identified discrepant values between the paired readers' grades. Manual measurement of morphologic features was deemed discrepant if there was more than 60 μm difference between vertical dimension measurements or 100 μm difference in horizontal dimension measurements. Caliper-based FTMH width grading was more stringent, and values were considered discrepant between readers if a disparity greater than 30 μm in horizontal dimension was observed. All graded scan pairs with discrepant data then were presented to a senior reader for arbitration. During the arbitration process, a senior reader reconciled all discrepancies between the initial reader pair and recorded the final arbitrated values. A senior reader additionally arbitrated all OCT scans for the presence of VMA, since this was an endpoint of the MIVI-TRUST program. Any finding or measurement that remained controversial after arbitration was forwarded to the Director of Grading for final decision. 
OCT Grading Reproducibility
From the 3695 scans obtained during the MIVI-TRUST program, 100 scans were selected randomly to determine Reading Center team grading reproducibility. The computerized scan selection algorithm was designed so that only one scan per individual study subject was included. The reproducibility testing was performed 5 weeks or more after initial grading to minimize the chance that readers would remember the initial evaluation. 22 Of the 100 scans selected initially, 10 contained FTMH. This number was below the minimal sample size (n = 36) determined to be necessary for meaningful statistical analysis. Therefore, 100 additional scans were selected randomly from the 587 scans that depicted FTMH to quantify effectively reproducibility of FTMH width measurements. 
For all reproducibility studies, a repeat grading team was defined as 2 readers chosen at random from the pool of 5 readers who performed the initial grading, matched with the same arbitrator who had performed the initial grading. For the reproducibility analysis, readers repeated grading of images in a number proportional to the number of images that they graded initially. Readers were masked to the results of the initial grading. After a second grading value was determined for a particular finding or measurement, this value was compared to the initial grade (Fig. 6). Reproducibility was determined for the categorical variables presence of VMA, broad versus focal width of vitreous adhesion, presence of FTMH, and presence of ERM. Quantitative grading reproducibility was analyzed for the minimum FTMH width measurement. 
Figure 6. 
 
OCT scan work-flow demonstrated for reproducibility analysis of hypothetical individual “Senior reader A.” Vitreous Adhesion Width = total vitreous adhesion width (horizontal dimension) between optic nerve and fovea on OCT image graded as either broad (>1500 μm) or focal (<1500 μm).
Figure 6. 
 
OCT scan work-flow demonstrated for reproducibility analysis of hypothetical individual “Senior reader A.” Vitreous Adhesion Width = total vitreous adhesion width (horizontal dimension) between optic nerve and fovea on OCT image graded as either broad (>1500 μm) or focal (<1500 μm).
Statistical Analysis
For categorical variables, the concordance, or percent agreement, was computed as a measure of reproducibility for repeated grading. The Kappa statistic, 23 a measure of agreement, also was computed. All analyses were performed with SAS 9.2 software (SAS, Cary, NC). 
For continuous values, paired differences between Reading Center teams were computed. The significance of the paired difference was assessed using a test of equality to the value 30.0 μm , the maximal limit of equivalence for FTMH width specified in the grading protocol. The intraclass correlation with confidence intervals (CI) was used to assess the agreement of continuous measurements. 
Results
We evaluated initial reader pair agreement, before arbitration on VMA presence or absence, and on VMA width, FTMH, and ERM. Of 100 randomly selected scans, during initial grading, reader agreement was 77% for scans with final grading as “VMA absent” (n = 22) and 80% for scans with final grading as “VMA present” (n = 78), respectively. During initial grading, the primary reader pair agreed on 78%, 91%, and 75% of scans independently evaluated in parallel for the parameters “broad versus focal width of vitreous adhesion,” “FTMH,” and “ERM,” respectively. Accordingly, arbitration was not performed for these parameters on the corresponding scans. In the remaining scans, the initial reader pairs disagreed with one another, and arbitration was required on 22%, 9%, and 25% of scans for the parameters “broad versus focal width of vitreous adhesion,” “FTMH,” and “ERM,” respectively. 
Reading Center team percent grading agreement was 97% for presence of VMA, 92% for broad versus focal width of vitreous adhesion, 95% for presence of FTMH, and 82% for presence of ERM. Reading Center team grading reproducibility was high for VMA, broad versus focal width of vitreous adhesion, FTMH, and ERM. Reading Center team qualitative grading reproducibility is shown in Table 1
Table 1. 
 
Team Reproducibility for Qualitative OCT Evaluation
Table 1. 
 
Team Reproducibility for Qualitative OCT Evaluation
Variable Agree Total Percent Agreement Kappa Statistic Kappa Statistic Lower 95% CL Kappa Statistic Upper 95% CL
VMA 97 100 97 0.91 0.81 1.00
Vitreous adhesion width (broad vs. focal) 92 100 92 0.87 0.78 0.95
FTMH 95 100 95 0.87 0.78 0.95
ERM 82 100 82 0.87 0.78 0.95
 Any retinal deformation at ERM 85 100 85 0.87 0.78 0.95
 Fovea deformed at ERM 82 100 82 0.87 0.78 0.95
 ERM at vitreomacular adhesion 80 100 80 0.61 0.46 0.75
For repeated Reading Center measurements of minimum FTMH width, the intraclass correlation was 0.89 (95% CI 0.85–0.93). For this measurement, the mean paired difference between grading team measurements was 34.4 μm (95% limits of agreement −149.5–218.2 μm, P = 0.643, Table 2). 
Table 2. 
 
Full Thickness Macular Hole Paired Differences
Table 2. 
 
Full Thickness Macular Hole Paired Differences
Statistic Initial Grading Repeat Grading Paired Difference
N 99 100 99
Mean (SD) 375.0 (212.2) 340.3 (221.2) 34.4* (93.8)
Min, median, max 69, 337, 1590 50, 296.5, 1559 −321, 12, 411
During repeat grading, the primary reader pair agreed on 63%, 92%, and 77% of scans evaluated independently in parallel for the parameters “broad versus focal width of vitreous adhesion,” “FTMH,” and “ERM,” respectively. Accordingly, arbitration was not performed for these parameters on the corresponding scans. In the remaining scans during repeat grading, the initial reader pairs disagreed with each other, and arbitration was required on 37%, 8%, and 23% of scans for the parameters “broad versus focal width of vitreous adhesion,” “FTMH,” and “ERM,” respectively. 
Discussion
In our study, we developed a scan protocol to evaluate vitreoretinal interface disorders, and studied the ability of Reading Center teams to grade reproducibly OCT images obtained with this protocol. We found that Reading Center teams were able to identify reproducibly VMA, associated pathologic features, such as FTMH and ERM, 16,24,25 and broad versus focal vitreous adhesion width in eyes with visually significant VMA at study baseline. In addition, individual reader pairs likewise demonstrated high levels of agreement when identifying the presence or absence of VMA and other vitreoretinal interface features. Thus, these methods are suitable to investigate a drug treatment effect in large prospective clinical trials, such as the MIVI TRUST program. Other studies have analyzed the reproducibility of OCT evaluation for diseases, such as neovascular age-related macular degeneration (NVAMD) 26 and central retinal vein occlusion (CRVO), 27 but to the best of our knowledge, this is the first to analyze reproducibility of a standardized grading protocol for evaluation of vitreoretinal interface disorders. 
Our method to characterize vitreoretinal interface morphology was modified from a study that described PVD evolution in 209 healthy eyes based on OCT and biomicroscopy. 19 While Uchino et al. detailed the natural history of PVD, our protocol sought to differentiate VMA from other types of posterior vitreous morphologies, since presence and subsequent resolution of VMA was a trial endpoint. We devised an OCT acquisition protocol to allow for efficient yet highly reproducible imaging that could be implemented readily in a standardized fashion across multiple clinical sites. In comparison, Uchino et al. performed at least six 7 mm images along the vertical and horizontal axes only to generate later composite images using commercial software for analysis. 19 Other investigators examining vitreoretinal interface disorders used radial line scans centered at the fovea, 12 or cross hair or radial line scans centered at the fovea. 20,28 Our protocol was designed to depict optimally the relationship among foveal, retinal, and optic nerve vitreous adhesions with a 10 mm line scan that passed through the optic nerve and fovea. Also, our protocol that required review of all 6 radial line images allowed readers to identify associated pathologic features that may not have been demonstrated on a horizontal or vertical line scan. The use of 10 mm radial line scans obviated the need for manual image compilation in our study. 
We likewise adapted the categorization scheme proposed by Uchino et al. to categorize comprehensively the various vitreous morphologies observed in this trial. Uchino's stage 0 was defined as absence of PVD, and stage 4 was defined as complete PVD out of range of OCT and verified by biomicroscopy. The ability of slit-lamp biomicroscopy to identify reliably vitreous separation remains to be determined. Regardless, since our study did not incorporate biomicroscopy, Uchino's stages 0 and 4 correlated to our “no visible vitreous” category. However, we included a “complete vitreous separation” category for those cases where complete macular PVD could be confirmed with OCT. 
The primary morphologic feature of interest during this trial, VMA, most closely corresponded to Uchino's stage 2, vitreous detachment with attachment at the optic nerve, fovea, and midperipheral retina. Uchino et al. did not describe our VMA subcategory of “vitreous adhesion to the fovea only.” In contrast, the two categories of partial vitreous separation with no visible separation elsewhere used in our study roughly correlated with Uchino's Stage 1 classification, defined as perifoveal PVD. Uchino's Stage 3 classification corresponded exactly to the “vitreous attached at the ONH and temporal macula with intervening separation” category from our study, though no mention was given to cases of “vitreous adhesion to the optic nerve only” as observed during our study. Finally, poor OCT scan quality precluding vitreomacular adhesion evaluation was not addressed by Uchino et al. 19 In anticipation of the several thousand scans to be generated during this clinical trial, we included a grading category for scans in which the quality was insufficient to interpret imaging parameters. 
Prior series detailing OCT grading protocols have used individual examiners, 27,29 and multiple grader pairs in parallel. 30 We chose a team-based grading approach with two primary readers and a senior reader who arbitrated scans to maximize grading consistency and to provide continued feedback between the senior reader and primary readers during the course of the study. In addition, this method allowed a senior reader to review more efficiently scans with subtle pathology, as opposed to more obvious abnormalities that can be graded effectively by primary readers without the need of senior reader arbitration. For example, the majority of grading performed in parallel by readers did not require arbitration in our study as demonstrated by the 8% to 37% range of arbitration frequencies. Our study demonstrates that team-based grading leads to increased levels of reproducibility when compared to agreement rates between independent primary readers. 
The protocol used in the ETDRS study for certain fundus photographs has most similarity to our team-based protocol for grading OCT scans. In the ETDRS study, baseline fundus photos were reviewed only by a pair of readers, and a subset of discrepancies were arbitrated by a senior reader. 31 Our protocol differed in that independent reading teams graded baseline and follow-up OCT scans, and a senior reader reconciled all discrepancies between the initial reader pairs. We believe that it was important to maintain this team analysis approach at each study visit, since consistent grading was important to evaluate reliably the primary study endpoint, which was a change in the OCT appearance of VMA from baseline to follow-up visits. 
Readers also generally were able to measure reproducibly minimum FTMH width. Repeated minimum FTMH width measurement resulted in narrower mean width, although this difference was not significantly different from the pre-specified 30 μm equivalence limit. The width differences from one evaluation to the next in our study may be due to measurement disparities among individual readers who made manual measurements at slightly different locations on the same radial line scans. Very small variations in Stratus on-screen caliper placement can cause a 30 μm or greater difference in width measurement. These measurement disparities were minimized by a grading protocol that required evaluation of all radial line images before macular hole width was determined for an individual patient with a FTMH. 
The primary question to be answered from the current report was whether the methods that we used to describe vitreoretinal interface abnormalities and the associated Reading Center team-based grading methodology could evaluate reproducibly the presence or absence of VMA and macular hole, and therefore determine reliably the key MIVI-TRUST endpoints, resolution of symptomatic VMA, and macular hole closure. At the time the MIVI-TRUST studies were initiated, it was not possible to obtain spectral domain (SD-OCT) data from most subjects enrolled in these studies, because SD-OCT was not available to the majority of study centers. Nonetheless, the key result from our study is that the methods used herein to characterize and analyze Stratus time domain OCT (TD-OCT) images are suitable to identify reproducibly initial presence and subsequent resolution of VMA and macular hole. 
Since the MIVI-TRUST studies were initiated, we anticipated that many clinicians would transition to SD-OCT in their practices. Accordingly, we performed an SD-OCT MIVI-TRUST substudy to compare morphologic and quantitative variables on eyes imaged with both TD-OCT and SD-OCT (data presented at the American Society of Retinal Specialists meeting, August, 2012, manuscript submitted for publication). The primary results of that substudy strongly support the validity of the data reported in the present report, that is there was no statistically significant difference in identification of primary study outcomes, VMA resolution, and macular hole closure, whether measured by TD-OCT or SD-OCT. A full description of that study is beyond the scope of our report; however, together with the SD-OCT data, the TD-OCT data demonstrated that for the sizeable number of practices that still use TD-OCT within and outside the United States, vitreomacular adhesion and macular hole diagnosis and resolution can be evaluated effectively and reproducibly with this OCT technology. 
The focus of our study was to determine whether a systematic method to analyze vitreoretinal interface disorders by OCT was reproducible and applicable to identify a drug-treatment effect when eyes are treated with vitreous pharmacolysis therapy. Since eyes were treated with enzymatic therapy or placebo, rather than with surgery, it was not possible to compare our data directly to observations obtained intraoperatively. Future studies to compare preoperative OCT findings to intraoperative clinical observations would be interesting in this regard. At the present time, as described above, the study results support the primary study objectives; that is the standardized, masked approach to evaluate OCT images of vitreoretinal interface disorders as used was reproducible, and suitable to investigate a drug treatment effect in large prospective multicenter clinical trials. 
There are limitations to our study. The relatively small reader pool used in this study pool precluded comparison of Reading Center teams, each comprised of completely different individuals. To address this limitation, all readers who participated in a reproducibility study were masked to the initial grading results, readers were selected randomly to create repeat grading teams, and grading reproducibility was tested several weeks after the initial scan was evaluated. Also, study images were analyzed at a single Reading Center. Accordingly, the reproducibility data may not apply to those obtained by other Reading Centers. Nonetheless, we believe that the principles used by the OCT scanning and analysis protocol in our study will be useful to ophthalmologists caring for patients with visually significant vitreomacular adhesions, and for future trials to investigate vitreoretinal interface disorders. 
References
Kohno T Sorgente N Ishibashi T Immunofluorescent studies of fibronectin and laminin in the human eye. Invest Ophthalmol Vis Sci . 1987;28:506–514. [PubMed]
Russell SR Shepherd JD Hageman GS. Distribution of glycoconjugates in the human retinal internal limiting membrane. Invest Ophthalmol Vis Sci . 1991;32:1986–1995. [PubMed]
Chen W Mo W Sun K Huang X Zhang YL Song HY. Microplasmin degrades fibronectin and laminin at vitreoretinal interface and outer retina during enzymatic vitrectomy. Curr Eye Res . 2009;34:1057–1064. [CrossRef] [PubMed]
Benz MS Packo KH Gonzalez V A placebo-controlled trial of microplasmin intravitreous injection to facilitate posterior vitreous detachment before vitrectomy. Ophthalmology . 117:791–797. [CrossRef] [PubMed]
de Smet MD Gandorfer A Stalmans P Microplasmin intravitreal administration in patients with vitreomacular traction scheduled for vitrectomy: the MIVI I trial. Ophthalmology . 2009;116:1349–1355. [CrossRef] [PubMed]
de Smet MD Valmaggia C Zarranz-Ventura J Willekens B. Microplasmin: ex vivo characterization of its activity in porcine vitreous. Invest Ophthalmol Vis Sci . 2009;50:814–819. [CrossRef] [PubMed]
Gandorfer A Rohleder M Sethi C Posterior vitreous detachment induced by microplasmin. Invest Ophthalmol Vis Sci . 2004;45:641–647. [CrossRef] [PubMed]
Koizumi H Spaide RF Fisher YL Freund KB Klancnik JM Jr Yannuzzi LA. Three-dimensional evaluation of vitreomacular traction and epiretinal membrane using spectral-domain optical coherence tomography. Am J Ophthalmol . 2008;145:509–517. [CrossRef] [PubMed]
Gallemore RP Jumper JM McCuen BW 2nd Jaffe GJ Postel EA Toth CA. Diagnosis of vitreoretinal adhesions in macular disease with optical coherence tomography. Retina . 2000;20:115–120. [CrossRef] [PubMed]
Hee MR Puliafito CA Wong C Optical coherence tomography of macular holes. Ophthalmology . 1995;102:748–756. [CrossRef] [PubMed]
Jumper JM Gallemore RP McCuen BW 2nd Toth CA. Features of macular hole closure in the early postoperative period using optical coherence tomography. Retina . 2000;20:232–237. [CrossRef] [PubMed]
Sonmez K Capone A Jr Trese MT Williams GA. Vitreomacular traction syndrome: impact of anatomical configuration on anatomical and visual outcomes. Retina . 2008;28:1207–1214. [CrossRef] [PubMed]
Massin P Duguid G Erginay A Haouchine B Gaudric A. Optical coherence tomography for evaluating diabetic macular edema before and after vitrectomy. Am J Ophthalmol . 2003;135:169–177. [CrossRef] [PubMed]
Kaiser PK Riemann CD Sears JE Lewis H. Macular traction detachment and diabetic macular edema associated with posterior hyaloidal traction. Am J Ophthalmol . 2001;131:44–49. [CrossRef] [PubMed]
Benhamou N Massin P Haouchine B Erginay A Gaudric A. Macular retinoschisis in highly myopic eyes. Am J Ophthalmol . 2002;133:794–800. [CrossRef] [PubMed]
Johnson MW. Posterior vitreous detachment: evolution and complications of its early stages. Am J Ophthalmol . 149:371–382. [CrossRef] [PubMed]
Krebs I Brannath W Glittenberg C Zeller F Sebag J Binder S. Posterior vitreomacular adhesion: a potential risk factor for exudative age-related macular degeneration? Am J Ophthalmol . 2007;144:741–746. [CrossRef] [PubMed]
Mojana F Cheng L Bartsch DU The role of abnormal vitreomacular adhesion in age-related macular degeneration: spectral optical coherence tomography and surgical results. Am J Ophthalmol . 2008;146:218–227. [CrossRef] [PubMed]
Uchino E Uemura A Ohba N. Initial stages of posterior vitreous detachment in healthy eyes of older persons evaluated by optical coherence tomography. Arch Ophthalmol . 2001;119:1475–1479. [CrossRef] [PubMed]
Johnson MW. Perifoveal vitreous detachment and its macular complications. Trans Am Ophthalmol Soc . 2005;103:537–567. [PubMed]
Stalmans P. Microplasmin to treat vitreomacular traction: results of the MIVI-007 phase III clinical trial. Presented at: 10th Euretina Congress; September 4, 2010; Paris, France.
Peto T. Reading centers of fundus imaging. J Ophthalmic Photog . 2010;32:47–49.
Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas . 1960;20:37–46. [CrossRef]
Michalewska Z Michalewski J Sikorski BL A study of macular hole formation by serial spectral optical coherence tomography. Clin Experiment Ophthalmol . 2009;37:373–383. [CrossRef] [PubMed]
Takahashi A Nagaoka T Ishiko S Kameyama D Yoshida A. Foveal anatomic changes in a progressing stage 1 macular hole documented by spectral-domain optical coherence tomography. Ophthalmology . 117:806–810. [CrossRef] [PubMed]
Krebs I Hagen S Brannath W Repeatability and reproducibility of retinal thickness measurements by optical coherence tomography in age-related macular degeneration. Ophthalmology . 2010;117:1577–1584. [CrossRef] [PubMed]
Domalpally A Blodi BA Scott IU The Standard Care vs Corticosteroid for Retinal Vein Occlusion (SCORE) study system for evaluation of optical coherence tomograms: SCORE study report 4. Arch Ophthalmol . 2009;127:1461–1467. [CrossRef] [PubMed]
Lee SJ Koh HJ. Effects of vitreomacular adhesion on anti-vascular endothelial growth factor treatment for exudative age-related macular degeneration. Ophthalmology . 2011;118:101–110. [CrossRef] [PubMed]
Krebs I Hagen S Brannath W Repeatability and reproducibility of retinal thickness measurements by optical coherence tomography in age-related macular degeneration. Ophthalmology . 2010;117:1577–1584. [CrossRef] [PubMed]
Zhang N Hoffmeyer GC Young ES Optical coherence tomography reader agreement in neovascular age-related macular degeneration. Am J Ophthalmol . 2007;144:37–44. [CrossRef] [PubMed]
Grading diabetic retinopathy from stereoscopic color fundus photographs –an extension of the modified Airlie House classification. ETDRS report number 10. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology . 1991;98:786–806. [CrossRef] [PubMed]
Footnotes
 Supported in part by the Heed Foundation.
Footnotes
 Disclosure: F.C. DeCroos, None; C.A. Toth, None; F.A. Folgar, None; S. Pakola, ThromboGenics (E); S.S. Stinnett, None; C.S. Heydary, None; R. Burns, None; G.J. Jaffe, None
Footnotes
 Current affiliation: *Amakem, Diepenbeek, Belgium.
Figure 1. 
 
Standardized OCT imaging protocols used to analyze the vitreoretinal interface during the MIVI 3 trial. White dashes: 6 radial line scans 6 mm in length were acquired using the FMTM and MTM protocols. White solid line: custom 10 mm horizontal and vertical line scans performed through the center of the ONH. Black dashes: custom 10 mm line scan through both the ONH and fovea angled at 5 degrees. All scans are high resolution (512 A-scans per tomographic line) except for those from the FMTM protocol (128 A-scans per tomographic line).
Figure 1. 
 
Standardized OCT imaging protocols used to analyze the vitreoretinal interface during the MIVI 3 trial. White dashes: 6 radial line scans 6 mm in length were acquired using the FMTM and MTM protocols. White solid line: custom 10 mm horizontal and vertical line scans performed through the center of the ONH. Black dashes: custom 10 mm line scan through both the ONH and fovea angled at 5 degrees. All scans are high resolution (512 A-scans per tomographic line) except for those from the FMTM protocol (128 A-scans per tomographic line).
Figure 2
 
Representative morphologic features noted on OCT images derived from the macular thickness map MTM protocol. (a) FTMH. (b) ERM.
Figure 2
 
Representative morphologic features noted on OCT images derived from the macular thickness map MTM protocol. (a) FTMH. (b) ERM.
Figure 3. 
 
Representative OCTs demonstrating caliper placement for each software-based measurements of FTMH width. (a) Maximum FTMH width at the level of the RPE. (b) Minimum FTMH width.
Figure 3. 
 
Representative OCTs demonstrating caliper placement for each software-based measurements of FTMH width. (a) Maximum FTMH width at the level of the RPE. (b) Minimum FTMH width.
Figure 4
 
OCT images depicting the types of vitreomacular adhesion observed during the MIVI 3 trial. (a, b) Examples of VMA. (a) Vitreous attachment at the ONH, and fovea and temporal macula with intervening separations. (b) Vitreous attachment at the fovea with separation on each side. (c) No visible vitreous from the ONH to the fovea with separation outside this zone (not considered VMA in study). (d) Vitreous attachment at the foveal edge of small FTMH and ONH with intervening separation and no visible vitreous separation temporal to the fovea (not considered VMA in study).
Figure 4
 
OCT images depicting the types of vitreomacular adhesion observed during the MIVI 3 trial. (a, b) Examples of VMA. (a) Vitreous attachment at the ONH, and fovea and temporal macula with intervening separations. (b) Vitreous attachment at the fovea with separation on each side. (c) No visible vitreous from the ONH to the fovea with separation outside this zone (not considered VMA in study). (d) Vitreous attachment at the foveal edge of small FTMH and ONH with intervening separation and no visible vitreous separation temporal to the fovea (not considered VMA in study).
Figure 5
 
OCT images depicting vitreous morphologies other than vitreomacular adhesion observed during the MIVI 3 trial. (a, b) No visible vitreous attachment or separation. (c) Vitreous attached at the ONH only.
Figure 5
 
OCT images depicting vitreous morphologies other than vitreomacular adhesion observed during the MIVI 3 trial. (a, b) No visible vitreous attachment or separation. (c) Vitreous attached at the ONH only.
Figure 6. 
 
OCT scan work-flow demonstrated for reproducibility analysis of hypothetical individual “Senior reader A.” Vitreous Adhesion Width = total vitreous adhesion width (horizontal dimension) between optic nerve and fovea on OCT image graded as either broad (>1500 μm) or focal (<1500 μm).
Figure 6. 
 
OCT scan work-flow demonstrated for reproducibility analysis of hypothetical individual “Senior reader A.” Vitreous Adhesion Width = total vitreous adhesion width (horizontal dimension) between optic nerve and fovea on OCT image graded as either broad (>1500 μm) or focal (<1500 μm).
Table 1. 
 
Team Reproducibility for Qualitative OCT Evaluation
Table 1. 
 
Team Reproducibility for Qualitative OCT Evaluation
Variable Agree Total Percent Agreement Kappa Statistic Kappa Statistic Lower 95% CL Kappa Statistic Upper 95% CL
VMA 97 100 97 0.91 0.81 1.00
Vitreous adhesion width (broad vs. focal) 92 100 92 0.87 0.78 0.95
FTMH 95 100 95 0.87 0.78 0.95
ERM 82 100 82 0.87 0.78 0.95
 Any retinal deformation at ERM 85 100 85 0.87 0.78 0.95
 Fovea deformed at ERM 82 100 82 0.87 0.78 0.95
 ERM at vitreomacular adhesion 80 100 80 0.61 0.46 0.75
Table 2. 
 
Full Thickness Macular Hole Paired Differences
Table 2. 
 
Full Thickness Macular Hole Paired Differences
Statistic Initial Grading Repeat Grading Paired Difference
N 99 100 99
Mean (SD) 375.0 (212.2) 340.3 (221.2) 34.4* (93.8)
Min, median, max 69, 337, 1590 50, 296.5, 1559 −321, 12, 411
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×