advertisement

---

Editors Selection IGR 21-2

Clinical Examination Methods: Retinal Layers in Glaucoma

Comment by Kouros Nouri-Mahdavi on:

78008 Diagnostic accuracy of macular ganglion cell-inner plexiform layer thickness for glaucoma detection in a population-based study: Comparison with optic nerve head imaging parameters, Koh V; Tham YC; Cheung CY et al., PLoS ONE, 2018; 13: e0199134

Find related abstracts

---

Koh et al. compared the diagnostic performance of macular and optic disc OCT parameters as measured with Cirrus HD-OCT to that of HRT3 for detection of glaucoma in the Singapore Chinese Eye study. The study provides us with important information regarding performance of the OCT or HRT3 in a population-based setting. There is a concern that case-control studies using patients and normal subjects enrolled in clinics could overestimate performance of diagnostic devices due to potential selection and/or ascertainment bias. The investigators found that the vertical cup-to-disc ratio (and inferior quadrant RNFL thickness) derived from SD-OCT performed better (AUC = 0.94) than the best macular GCIPL measure (minimal GCIPL, AUC = 0.89) or HRT3's vertical CDR (AUC = 0.86) based on area under ROC curves or sensitivity at 85% specificity (sensitivities of 89%, 61%, and 65%, respectively).

There is a concern that case-control studies using patients and normal subjects enrolled in clinics could overestimate performance of diagnostic devices due to potential selection and/ or ascertainment bias

The results are encouraging and confirm the potential role of SD-OCT optic disc/RNFL measures for glaucoma screening or detection of glaucoma under other settings. It is not unexpected to see OCT-derived CDR or RNFL thickness measures outperform macular parameters for detection of glaucoma as the former measures are better reflective of the entire RGC complement of an eye than macular measures. However, the results should be interpreted taking into account of a few caveats. It seems that diagnosis of glaucoma, or lack thereof, was established in real-time by clinical examination of the optic disc by a single clinician and therefore, one cannot rule out possible ascertainment bias that could have actually diluted the performance of OCT measures. The mean deviation of the glaucoma group was about –9 dB and therefore, the glaucoma subgroup had moderate disease on average. This could, conversely, have led to a more optimistic impression of OCT or HRT3's performance. The average axial length was 23.9 mm in both glaucoma and normal groups, on the low side for an Asian population, and thus limiting the generalizability of the findings to myopic individuals. It would have been interesting to see the proportion of eyes detected by each of the best-performing measures in a Venn diagram to see how complementary the corresponding information would be. Also, an ROC curve for a linear combination of the best-performing parameters from each modality would have been very useful to confirm or refute the utility of using combined parameters from different diagnostic domains.

I would like to commend the authors for providing much needed data to better clarify the potential utility of OCT devices for screening in real-world setting.

---

• ← Previous
• Next →

---