Bony nasolacrimal canal morphometry and sinonasal anatomy in primary acquired nasolacrimal duct obstruction: a retrospective comparative computed tomography study.

PURPOSE: To compare bony nasolacrimal canal (BNLC) morphometry and adjacent sinonasal measurements between patients with primary acquired nasolacrimal duct obstruction (PANDO) and healthy controls, and to determine whether these anatomical characteristics are associated with PANDO. METHODS: In this retrospective, cross-sectional study, computed tomography (CT) images from 86 PANDO patients and 88 healthy controls (n = 174) were analysed. Nine parameters were measured using a standardised protocol: proximal area, distal area, BNLC centre-nasal base angle, canal length, BNLC long axis-midline angle, inferior meatus width, inferior turbinate thickness, nasal width, and nasal height. In patients with unilateral PANDO, the affected and unaffected canals were additionally compared with paired tests. Group differences were assessed with independent-samples t-tests or Mann-Whitney U tests, two-way ANOVA (PANDO status × sex), binary logistic regression, and ROC curve analysis. The Benjamini-Hochberg procedure controlled the false discovery rate across the nine morphometric comparisons. RESULTS: PANDO patients were older than controls (42.26 ± 15.46 vs. 31.18 ± 9.84 years; p < 0.001; Cohen's d = 0.857). Canal length (16.55 ± 3.78 vs. 17.91 ± 3.50 mm) and the BNLC long axis-midline angle (7.96 ± 4.32° vs. 9.48 ± 4.17°) were both lower in PANDO patients on univariate testing (raw p = 0.021 and p = 0.007), but neither difference remained significant after Benjamini-Hochberg correction (q = 0.095 and q = 0.063). No other morphometric parameter differed between groups. In logistic regression, age (OR = 1.074; 95% CI: 1.043-1.107) and a smaller BNLC long axis-midline angle (OR = 0.894; 95% CI: 0.824-0.971) were independently associated with PANDO (Nagelkerke R2 = 0.306); canal length and sex were not. The observed between-group difference in the midline angle (1.52°) was smaller than its estimated minimal detectable change (MDC95 ≈ 3.90°), indicating that this difference cannot be distinguished from measurement error with confidence. CONCLUSION: A smaller BNLC long axis-midline angle and older age were associated with PANDO status in this cohort, but the angular finding did not survive multiplicity correction and fell below the threshold of reliably detectable change. These results are hypothesis-generating rather than confirmatory. Prospective, multicentre studies with three-dimensional imaging and pre-registered analysis plans are needed to establish whether canal orientation contributes to PANDO pathogenesis.