Defining Clinically Significant Vitreous Floaters with Objective Quantitative Measures of Vitreous Structure and Visual Function

Abstract

Purpose : To define clinically significant vitreous floaters by determining objective structural and functional characteristics which discriminate between floater patients desiring vitrectomy, and those not sufficiently bothered to undergo surgery.

Methods : 64 eyes from 45 subjects (32 men, 13 women; 57 +/- 15 years) complaining of floaters were grouped by desire for vitrectomy (PPV; N=24) or observation (N=40). The etiology of floaters was posterior vitreous detachment in 40 (62.5%) and myopia in 24 (37.5%). Vitreous structure was evaluated using quantitative ultrasonography (QUS) to measure vitreous echodensities [Mamou et al. IOVS 2015:56;1511-7]. Visual function was measured with Snellen visual acuity (VA) and contrast sensitivity (CS; %Weber: Freiburg Acuity Contrast Testing). The predictive value of each measure was expressed as the area under the receiver operating characteristic (ROC) curve (AUC: range = 0.5 (no predictive value) to 1.0).

Results : Structurally, QUS in the observation group was 547 vs. 774 units in vitrectomy eyes (42% greater echodensity; P<0.001). Pre-macular QUS had an AUC of 0.770. Functionally, decimal VA for the observation group averaged 0.76 vs. 0.66 in the vitrectomy group (P<0.02). VA had an AUC of 0.633, the lowest of all the parameters. CS for the observation group averaged 2.55%W vs. 3.93%W in the vitrectomy group (54% worse; P<0.001). CS had an AUC of 0.813.
Weighting and combining CS with QUS parameters linearly (0.720 * CS + 0.012 * QUS) maximized AUC at 0.830. This combined parameter had a sensitivity of 91.7% and a specificity of 75.0%. The accuracy of the ROC curve was 81.3%.

Conclusions : Structural measures by quantitative ultrasonography (QUS) as well as functional measures of VA and contrast sensitivity (CS) can be used to objectively evaluate patients with floaters. A patient's desire to undergo vitrectomy is best predicted by combining QUS and CS, and least predicted by VA. These indices may thus be useful to objectively identify surgical candidates on the basis of clinically significant floaters as defined by structure and function.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

Fig 1. The ROC curve represents the sensitivity of a clinical measure as a function of its specificity. As seen above, VA had the lowest AUC. QUS can predict a patient’s desire for surgery, but CS had the highest AUC of any single parameter. Combining CS and QUS was the best predictor (highest AUC).