Associations between quantitative ocular surface features and CT morphology of the levator palpebrae superioris complex in patients with thyroid eye disease and upper eyelid retraction: a cross-sectional study.
Authors
Affiliations (2)
Affiliations (2)
- The Second People's Hospital of Chun'an County, Chun'an County, China.
- National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, China.
Abstract
This study aimed to investigate the associations between artificial intelligence (AI)-quantified ocular surface parameters and computed tomography (CT)-based morphology of the levator palpebrae superioris complex (LPSC) in patients with Thyroid eye disease (TED) and upper eyelid retraction. This retrospective cross-sectional study included 67 patients with predominantly isolated TED-related upper eyelid retraction (89 eyes) between May 2023 and October 2024. AI models were used to quantify meibomian gland vagueness, meibomian gland density, and tear meniscus height (TMH). LPSC thickness and area were measured on orbital CT images. Correlation analyses, partial correlation analyses, generalized estimating equations (GEE), and exploratory parallel mediation analyses were performed. Meibomian gland vagueness was negatively correlated with anterior LPSC area (<i>r</i> = -0.333, <i>p</i> = 0.002), middle LPSC thickness (<i>r</i> = -0.287, <i>p</i> = 0.008), and middle LPSC area (<i>r</i> = -0.352, <i>p</i> = 0.001). After adjustment for clinical activity score, proptosis, and upper lid marginal reflex distance (MRD1), these correlations remained significant for anterior LPSC area (<i>r</i> = -0.372, <i>p</i> < 0.001), middle LPSC thickness (<i>r</i> = -0.297, <i>p</i> = 0.009), and middle LPSC area (<i>r</i> = -0.369, <i>p</i> < 0.001). GEE analysis identified anterior LPSC area (<i>β</i> = -0.168, 95% CI: -0.291 to -0.045, <i>p</i> = 0.008) and middle LPSC area (<i>β</i> = -0.144, 95% CI: -0.236 to -0.053, <i>p</i> = 0.002) as independent factors associated with meibomian gland vagueness. TMH was positively correlated with middle LPSC thickness (<i>r<sub>s</sub></i> = 0.254, <i>p</i> = 0.017) and middle LPSC area (<i>r<sub>s</sub></i> = 0.348, <i>p</i> < 0.001) in bivariate analyses, but these associations were no longer significant after adjustment. Proptosis (<i>β</i> = 0.016, 95% CI: 0.009 to 0.024, <i>p</i> < 0.001) and MRD1 (<i>β</i> = -0.025, 95% CI: -0.040 to -0.010, <i>p</i> = 0.001) were independently associated with TMH. Exploratory mediation analysis showed significant indirect effects through proptosis for the associations of middle LPSC thickness with TMH and middle LPSC area with TMH. In this selected cohort of patients, LPSC enlargement was associated with reduced meibomian gland vagueness. The relationship between LPSC enlargement and increased TMH appeared to be indirect and may involve proptosis.