Response to the Letter to the Editor: Diagnostic Capacity and Correlation of Optic Nerve Colorimetry With Structural Parameters in Primary Open Angle Glaucoma

In Reply: We appreciate the opportunity to respond to the letter submitted by Prof Manuel González de la Rosa regarding our recently published article in the Journal of Glaucoma.1 We value his observations and acknowledge the importance of rigorous methodological analysis as a foundation for scientific progress. He identified a relevant methodological consideration, specifically a potential incorporation bias. In our study, glaucoma diagnosis was confirmed according to the standard definition, based on structural damage (optic nerve/pRNFL assessment) and/or functional damage (VF). Peripapillary retinal nerve fiber layer (pRNFL) thickness was assessed by OCT. Specifically, the relevant inclusion criteria were: Clinical features consistent with optic nerve damage. Confirmed abnormal visual field (VF) test (mean defect > 2 dB) and/or decreased pRNFL thickness. Open angle on gonioscopy. We wish to clarify that, in the original definition of primary open angle glaucoma (POAG), the conjunction “or” was inadvertently omitted—this was a grammatical oversight. This manuscript forms part of a doctoral thesis already presented.2 The possibility of an incorporation bias cannot be ruled out, as optic nerve head damage—specifically OCT-measured pRNFL thickness—was used as part of the diagnostic criteria in our study. OCT and Laguna ONhE were later compared, and this overlap could have made the structural test performance appear more favorable than it truly is. As a result, the diagnostic agreement observed between structural parameters and the Laguna ONhE tool should be interpreted with caution. While this overlap does not invalidate our results, it means that structural parameters were not entirely independent of the diagnostic reference. We agree with the methodological concerns raised by Prof González de la Rosa and recognize the value of the references he provided, including his group’s work highlighting methodological risks in diagnostic tool validation. We also appreciate his offer to provide free access to the latest version of the Laguna ONhE software for research purposes, which will facilitate future independent comparisons and collaborative studies. For transparency, we reviewed our data set of 112 patients with POAG, obtaining 23.2% of patients were included based solely on structural criteria (abnormal OCT) despite having normal visual fields. In addition, 6 patients presented functional damage on VF before structural loss was detectable by OCT (Table 1). TABLE 1 - Distribution of POAG Subjects According to VF and OCT Findings Diagnostic basis Patients (n) Percentage (%) Normal VF (MD ≤ 2 dB) + abnormal OCT 26 23.20 Abnormal VF (MD > 2 dB) + abnormal OCT 75 66.90 Abnormal VF (MD > 2 dB) + normal OCT 6 5.35 VF data unavailable 5 4.46 POAG indicates primary open angle glaucoma; VF, visual field; MD, mean defect; OCT, optical coherence tomography. We performed a new analysis on the 81 patients after excluding those diagnosed exclusively based on abnormal OCT findings and normal VF. The results, summarized in Table 2, were not substantially different from the original analysis of all 112 POAG patients. TABLE 2 - Comparison of AUCs From the new Analysis* and the Original Analysis GDF POAG vs AUC (95% CI); n=81 P paired AUC (95% CI); n=112 P paired RNFLT 3.5 mm G 0.942 (0.903–0.981) 0.0044 0.920 (0.880–0.960) 0.0007 BMO-MRW G 0.907 (0.857–0.958) 0.0156 0.898 (0.854–0.942) 0.0011 GCL T2 0.910 (0.868–0.952) 0.0429 0.883 (0.838–0.928) 0.0023 GDF 0.823 (0.758–0.888) — 0.801 (0.740–0.861) — *The new analysis on the 81 patients after excluding those diagnosed exclusively based on abnormal OCT findings and normal VF.AUC indicates areas under the curves; POAG, primary open angle glaucoma; GDF, glaucoma discriminant function; RNFLT 3.5 mm, peripapillary retinal nerve fiber layer in diameter 3.5 mm by the Glaucoma Module Premium Edition program; BMO-MRW: Bruch’s membrane opening-minimum rim width; GCL, macular ganglion cell layer; G, global; T2, outer temporal. These findings confirm that, even after excluding patients diagnosed solely based on OCT findings, structural tests (RNFLT, BMO-MRW, and GCL) maintained a high diagnostic performance, consistently outperforming the functional GDF index. This supports the robustness of our conclusions while highlighting the need to acknowledge the potential influence of incorporation bias when interpreting the results.