Fundus Photos More Reliable than OCT for Recognizing AMD Progression, Study Says

A study in Ophthalmic and Physiological Optics reveals challenges in using OCT biomarkers to predict AMD progression. While OCT shows promise, fundus photography proved more accurate (77.7%). The study indicates that combining at least three OCT biomarkers significantly improves prognostic performance (91.0%), highlighting the need for automated OCT interpretation to enhance clinical efficiency and patient care. Photo: Anna Bedwell, OD. Click image to enlarge. The ability to discern which patients with age-related macular degeneration (AMD) are at higher risk of progressing to late AMD is a distinction that can make care more targeted and timely. OCT has contributed greatly to these efforts by allowing the detection and tracking of key biomarkers; however, the accuracy of these predictions has been called into question. A new study published in Ophthalmic and Physiological Optics evaluated whether OCT biomarkers more accurately predict AMD progression compared to traditional color fundus photography, the results of which suggest that the complex interpretability of high-risk OCT biomarkers make it an ongoing challenge for practice integration.The retrospective study included 78 single eyes with intermediate AMD. Participants were an average of 71.8 years of age, 56% were women and 50% of white race/ethnicity. The groups were propensity-score matched by age and sex between converters and non-converters to late AMD. Researchers used 10 specific OCT biomarkers and two fundus photo biomarkers (large drusen and pigmentary abnormality), which were independently graded by three researchers. Statistical analyses were then performed to evaluate biomarker prevalence, reliability of grading, time to conversion and their ability to predict late AMD.According to the results, the adjusted risk was highest for OCT-detected nascent geographic atrophy (GA), shallow irregular RPE elevations, drusenoid pigment epithelium detachment and RPE reflective abnormality (odds ratios, 6.66 to 28.27).Fundus-detected pigmentary abnormalities demonstrated the highest individual prognostic accuracy (77.7% AUC), with excellent sensitivity (92.3%) but moderate specificity (63.1%). Adding at least three OCT biomarkers was required to improve prognostic performance significantly (91.0%), and at least eight additional biomarkers to yield both excellent sensitivity (94.6%) and specificity (90.8%). The study authors wrote in their paper that pigmentary abnormality detected by retinal photography “remains a mainstay of clinical AMD prognostication, likely due to its higher prevalence and interpretability” and cautioned that “integrating OCT biomarkers into clinical prognostic models is promising but complex and may require automated identification to aid efficiency.”They also noted that, although biomarkers such as nascent GA and shallow irregular RPE elevation were strongly associated with progression to late AMD, their relatively lower prevalence likely limited their prognostic utility. “As such, high-risk but less common biomarkers may not be as useful in future imaging-based prognostic models unless supported by more common biomarkers” such as reticular pseudodrusen, they wrote. “Large drusen also lacked discriminatory power, though likely because they were nearly ubiquitous in eyes with intermediate AMD. Meanwhile, pigmentary abnormality emerged as the single most useful predictor of disease progression, combining moderate prevalence in both converter and non-converter eyes with strong discriminatory performance.”The reliability of the grading personnel to identify biomarkers on OCT underscores the need for biomarkers that non-specialists can identify objectively and automatically, possibly with the use of artificial intelligence, the authors wrote. Some weaknesses mentioned in the study include its relatively small sample size from a single eyecare center, and the use of one type of OCT device, which limits how broadly the findings can be applied. Additionally, internal validation through data splitting or cross-validation was not performed due to the constraints of the sample size, possibly increasing the risk of overfitting during the biomarker selection. “Finally,” continued the authors, “manual identification of biomarkers is time intensive and subject to inter-grader variability, although automated image interpretation is anticipated to become increasingly widespread.” These approaches will continue to require validation against expert consensus as the reference standard to ensure ongoing relevance, they noted.Ultimately, the authors say this study confirmed the reliability of fundus photography as a crucial tool in predicting AMD progression, and more common use of OCT biomarkers will require some form of automated identification in order to improve its efficiency in clinical settings.Click here for the journal source. Ansari SM, Nguyen T, Khankan R, et al. Meibomian gland characteristics in children: a narrative review. Ophthalmic Physiol Opt. July 2, 2025. [Epub ahead of print].