Cluster Analysis Models Demonstrate Superiority to Pointwise Methods for Predicting Visual Field

Cluster analysis-based models which incorporate age correction and consider fovea to optic disc tilt demonstrate superior performance in predicting visual field (VF) results compared with pointwise methods in both eyes with glaucoma and healthy eyes, according to research published in Ophthalmic and Physiological Optics. However, low sensitivity and poor performance at the peripheral macula indicate that optical coherence tomography (OCT) alone may be insufficient to accurately predict visual function across the entire macula. 

Researchers collected data from 77 participants consisting of individuals with glaucoma (n=37) and healthy controls (n=40). They compared data from these prospectively recruited participants with data obtained from a normative cohort of 493 retrospectively recruited healthy participants. All participants underwent at least 1 comprehensive eye examination including slit-lamp evaluation, intraocular pressure measurement, dilated fundus examination, OCT imaging of the macula and peripapillary retinal nerve fiber layer (RNFL), and 24-2 VF testing. The team used this data to create a cluster and pointwise analysis. 

All participants with glaucoma underwent selective laser trabeculoplasty, peripheral iridotomy or were prescribed topical therapy at the time of recruitment. The majority were classified as having early glaucoma. 

Investigators reported that participants with glaucoma were significantly older and more hyperopic than control individuals (P <.0001 and P =.02, respectively), and the fovea to optic disc tilts were significantly less in the healthy testing cohort (P =.008). The mean and pattern standard deviations were also significantly worse in patients with glaucoma (P <.0001 for both).  

Investigators determined that clustered models demonstrated globally low sensitivity, but high specificity in patients with glaucoma (0.28–0.53 and 0.77–0.91, respectively), and high specificity in control individuals (0.91–0.98). Clustered models showed similar sensitivities and superior specificities compared with pointwise methods (0.41– 0.65 and 0.71–0.98, respectively). There were significant differences in accuracy between clusters, with relatively poor accuracy at peripheral macular locations (P < .0001 for all comparisons), according to the report.

“Relatively low sensitivity and poorer performance at the peripheral macula indicate that OCT in isolation may be insufficient to predict visual function across the macula accurately,” according to the researchers. “With modifications to criteria for abnormality, the concepts suggested by the described normative models may guide prioritization of VF assessment requirements, with the potential to limit excessive VF testing.”

Study limitations include a small sample size, low number of participants with macular VF defects, and inability  to test GCIPL thickness cut-offs.  

Reference

Tong J, Alonso-Caneiro D, Kalloniatis M, Zangerl B. Prediction of visual field defects from macular optical coherence tomography in glaucoma using cluster analysis. Ophthalmic Physiol Opt. Published online May 22. doi:10.1111/opo.12997