From the editors of Review of Ophthalmology:
MARCH IS WORKPLACE EYE WELLNESS MONTH
In this issue: (click heading to view article)
Deep Learning Predicts OCT Measures of Diabetic Macular Thickening
Researchers developed deep learning (DL) models for the automatic detection of optical coherence tomography measures of diabetic macular thickening from color fundus photographs (CFPs).
The retrospective analysis included 17,997 CFPs and their associated OCT measurements from the Phase III RIDE/RISE diabetic macular edema studies. Researchers applied DL with transfer-learning cascades on CFPs to predict time-domain OCT-equivalent measures of macular thickening, including central subfield thickness and central foveal thickness. They defined MT by using two OCT cutoff points: 250 μm and 400 μm. And they developed a DL regression model to directly quantify the actual CST and CFT from CFPs. The following were some of the findings:
• The best DL model was able to predict CST ≥250 μm with an area under the curve of 0.97 (CI, 0.89 to 1) and CFT ≥250 μm with an area under the curve of 0.91 (CI, 0.76 to 0.99).
• The best DL model was able to predict CST ≥400 μm with an AUC of 0.94 (CI, 0.82 to 1.00) and CFT ≥400 μm with an AUC of 0.96 (CI, 0.88 to 1).
• The best deep convolutional neural network regression model to quantify CST had an R2 of 0.74 (CI, 0.49 to 0.91) and to quantify CFT had an R2 of 0.54 (CI, 0.20 to 0.87).
• The performance of the DL models declined when CFPs were of poor quality or contained laser scars.
Researchers concluded that DL was capable of predicting key quantitative TD-OCT measurements related to MT from CFPs. They added that DL models presented in the study could enhance the efficiency of DME diagnosis in tele-ophthalmology programs, which might promote better visual outcomes. However, researchers wrote that future research would be needed to validate DL algorithms for MT in the real world.
SOURCE: Arcadu F, Benmansour F, Maunz A, et al. Deep learning predicts OCT measures of diabetic macular thickening from color fundus photographs. IOVS 2019;60:852-7.
Characteristics of Type 3 Neovascularization Lesions
Investigators evaluated the incidence of multifocal lesions and distribution of lesion location in type 3 neovascularization, as part of a retrospective, observational study including 148 eyes of 148 people diagnosed with type 3 neovascularization.
Investigators counted the number of type 3 neovascularization lesions and estimated the incidence of multiple lesions in an eye. In addition, they estimated the distance from the fovea to the lesion, and the geographic location of the lesion. Investigators also compared pseudodrusen incidence between eyes with and without multifocal lesions. Here were some of the findings:
• Investigators reported a total of 169 type 3 neovascularization lesions.
• They found a single lesion in 130 eyes (87.8 percent), and two or three multifocal lesions in the remaining 18 eyes (12.2 percent).
• The mean distance from the fovea to the lesion was 898.8 ±324.9 µm.
• The distribution of lesion locations exhibited a fovea-sparing pattern.
• No lesions were located within 200 µm of the fovea; 20 lesions (11.8 percent) were located >200 and ≤500 µm away from the fovea; 89 lesions (52.7 percent) were located >500 and ≤1,000 µm away from the fovea; and 60 lesions (35.5 percent) were located >1,000 µm away from the fovea.
• Pseudodrusen incidence was significantly higher in eyes with multifocal lesions (p=0.024).
Investigators reported two or more multifocal lesions in 12.2 percent of eyes with type 3 neovascularization, and they noted that pseudodrusen incidence was higher in eyes with multifocal lesions. In addition, investigators found that lesion distribution exhibited a fovea-sparing pattern—characteristics that they said might be associated with the distinct pathophysiology of type 3 neovascularization.
SOURCE: Kim JH, Chang Y, Kim JW, et al. Characteristics of type 3 neovascularization lesions. Retina 2019; Feb. 28. [Epub ahead of print].
Efficacy & Safety of Bleb Needle Revision with & without 5-FU for Failing Trabeculectomy Bleb
Scientists investigated the efficacy and safety of bleb needle revision with adjunctive 5-fluorouracil subconjunctival injection in individuals with failing trabeculectomy blebs. They performed a retrospective chart review of 106 subjects (113 eyes) who had undergone trabeculectomy followed by needle revision and had been followed for at least six months.
The subjects were divided into two groups according to whether they had also undergone concurrent adjunctive 5-FU subconjunctival injection. Scientists used two definitions of failure, irrespective of IOP-lowering medication use: 1) IOP>21 mmHg or >85 percent of preoperative IOP or reoperation for glaucoma; and 2) IOP >18 mmHg or >80 percent of preoperative IOP or reoperation for glaucoma.
A total of 73 subjects (78 eyes) underwent the needle revision procedure, and 33 subjects (35 eyes) underwent needle revision with subconjunctival 5-FU injections. Here were some of the findings:
• No significant adverse events were reported for either group, and no significant IOP differences were found in any observational periods between them (p=0.867).
• No statistically significant differences were found in the total number of IOP-lowering medications between the groups (p=0.199).
• The success rate at 24 months after the first needle revision was 45.6 percent (PI, 33.7 to 56.7) in the simple revision group and 44.6 percent (PI, 27.7 to 60.2) in the 5-FU needle revision group by failure 1.
• The success rate at 24 months after the first needle revision was 36.3 percent (PI, 25.4 to 47.2) in the simple revision group and 32.9 percent (PI; 18 to 48.6) in the 5-FU needle revision group by failure 2.
• A greater total number of needle revisions was the significant risk factor for needle revision failure (HR=2.25 and p=0.001 according to failure 2).
Scientists found that bleb needle revision with 5-FU subconjunctival injection was a safe procedure for a failing trabeculectomy bleb and that bleb needle revision with and without 5-FU were equally likely to succeed.
SOURCE: Kim J-S, Kim HJ, Na KI, et al. Comparison of efficacy and safety of bleb needle revision with and without 5-fluorouracil for failing trabeculectomy bleb. J Glaucoma 2019; March 4. [Epub ahead of print].
Corneal Higher Order Aberrations in DMEK vs. UT-DSAEK in DETECT Study
Researchers compared corneal higher order aberrations (HOAs) following ultrathin Descemet’s stripping automated endothelial keratoplasty and Descemet’s membrane endothelial keratoplasty, as part of an outcome-masked, randomized controlled clinical trial called DETECT.
Participants included individuals with damaged or diseased endothelium from Fuchs’ endothelial dystrophy or pseudophakic bullous keratopathy who were considered good candidates for DMEK or UT-DSAEK.
Corneal anterior and posterior surface HOAs were measured with Scheimpflug imaging before surgery and at three, six and 12 months postoperatively. Zernike orders 3-8 were calculated at 4- and 6-mm-diameter optical zones (OZs). HOAs after UT-DSAEK and DMEK were compared, and correlation was performed between best spectacle-corrected visual acuity and HOAs at each time point.
Main outcome measures included higher order aberrations of the anterior and posterior cornea, expressed as the root mean square deviation from a best fit sphere reference surface. Here were some of the findings:
• At three, six and 12 months after surgery, the posterior corneal surface had significantly less coma (p≤0.003) and total HOAs (p≤0.001) in DMEK compared with UT-DSAEK (4- and 6-mm OZ).
• Posterior trefoil (p≤0.034), secondary astigmatism (p≤0.042) and tetrafoil (p≤0.045) were lower in DMEK than UT-DSAEK at three, six or 12 months (either 4- or 6-mm OZs).
• No significant differences were found in anterior surface HOAs between DMEK and UT-DSAEK at any time point.
• Compared with baseline, total posterior HOAs significantly increased (p≤0.036) with UT-DSAEK at three, six and 12 months, in contrast to DMEK where they significantly decreased (p≤0.044) at six and 12 months (4- and/or 6-mm OZ).
• At six months (p≤0.635) and 12 months (p≤0.001), posterior corneal total HOAs correlated moderately with BSCVA (4- and 6-mm OZ).
• Researchers found no moderate or strong correlations between anterior or combined corneal surface HOAs at any time points after surgery.
Researchers determined that DMEK resulted in less posterior corneal HOAs compared with UT-DSAEK. As well, they wrote that DMEK decreased and UT-DSAEK increased posterior corneal HOAs compared with pre-surgical values. Furthermore, researchers found that total posterior corneal HOAs correlated moderately with six- and 12-month postoperative visual acuity and that they might partially account for better visual acuity observed after DMEK.
SOURCE: Duggan MJ, Rose-Nussbaumer J, Lin CC, et al. Corneal higher-order aberrations in DMEK versus UT-DSAEK in DETECT: A randomized clinical trial. Ophthalmology 2019; Feb. 15. [Epub ahead of print].
Roche and Genentech Initiate Two Phase III Studies in Wet AMD for Faricimab; Study Shows Evidence AI Can Detect DME Severity
Roche and Genentech initiated two Phase III clinical trials investigating the bispecific molecule faricimab, which simultaneously binds to and inactivates Angiopoietin-2 and vascular endothelial growth factor A, to treat retinal eye diseases. By targeting Ang-2 and VEGF-A, faricimab may lead to sustained efficacy at longer treatment intervals, thereby improving vision outcomes for patients. The identically designed, multicenter, randomized, double-masked, active comparator-controlled Phase III TENAYA and LUCERNE studies will evaluate the efficacy, safety and durability of faricimab compared with aflibercept for the treatment of wet age-related macular degeneration. The primary endpoint of each study is the change in best-corrected visual acuity at week 48 from baseline.
Read more about TENAYA
Read more about LUCERNE
In addition, a study published as part of Roche/Genentech’s Ophthalmology Personalized Healthcare initiative in Investigative Ophthalmology and Visual Science suggests that artificial intelligence could be used to provide widespread, cost-effective eye screenings via telemedicine to assist ophthalmologists in improving vision outcomes for diabetics who may not be getting regular eye exams. The research assessed how deep learning can automatically view color fundus photographs to accurately detect DME and determine its severity. Read more.
Horizon Pharma Phase III Trial Evaluating Teprotumumab Meets Endpoints
Horizon Pharma announced its Phase III confirmatory trial evaluating teprotumumab for the treatment of active thyroid eye disease met its primary endpoint. The study found that more individuals treated with teprotumumab compared with placebo had meaningful improvements in proptosis: 82.9 percent of teprotumumab patients compared with 9.5 percent of placebo subjects achieved the primary endpoint of a 2 mm or greater reduction in proptosis (p<0.001). Proptosis is the main cause of morbidity in thyroid eye disease, the company says. All secondary endpoints were also met and the safety profile was consistent with the Phase II study of teprotumumab in thyroid eye disease. Read more.
Biogen to Acquire Nightstar Therapeutics
Biogen entered into an agreement to acquire Nightstar Therapeutics (NST), a London-based, clinical-stage gene therapy company focused on adeno-associated virus treatments for inherited retinal disorders. NST’s lead asset is NSR-REP1 for the treatment of choroideremia, composed of an AAV vector administered by subretinal injection that provides a functioning CHM gene and expression of the REP-1 protein to restore membrane trafficking. NSR-REP1 is being evaluated in the ongoing Phase III STAR trial. NST’s second clinical program is NSR-RPGR for the treatment of X-linked retinitis pigmentosa. Biogen expects to complete the acquisition by mid-2019. Read more.
Review of Ophthalmology® Online is published by the Review Group, a Division of Jobson Medical Information LLC (JMI), 11 Campus Boulevard, Newtown Square, PA 19073.
To subscribe to other JMI newsletters or to manage your subscription, click here.
To change your email address, reply to this email. Write "change of address" in the subject line. Make sure to provide us with your old and new address.
To ensure delivery, please be sure to add firstname.lastname@example.org to your address book or safe senders list.
Click here if you do not want to receive future emails from Review of Ophthalmology Online.
Advertising: For information on advertising in this e-mail newsletter or other creative advertising opportunities with Review of Ophthalmology, please contact sales managers James Henne or Michele Barrett.
News: To submit news or contact the editor, send an e-mail, or FAX your news to 610.492.1049