Diagnostic Accuracy of a Humphrey 24-2C Protocol for Simultaneous Parafoveal and Pericentral Hydroxychloroquine Retinopathy Screening

Summary

Hydroxychloroquine toxicity: New Humphrey 24-2C protocol is a reliable single-test solution to screen for both parafoveal and pericentral toxicity. Comparable to 10-2 + 30-2 in a single test.

Abstract

Purpose: To evaluate the diagnostic performance of the Humphrey 24-2C visual field (VF) protocol compared with the standard 10-2 and 30-2 strategies for detecting parafoveal and pericentral hydroxychloroquine (HCQ) retinopathy.

Design: Retrospective cross-sectional diagnostic validity assessment.

Participants: In total, 1081 patients (2104 eyes) receiving HCQ therapy without confounding ocular pathology were screened for retinopathy at a tertiary center between January 2021 and April 2025. The cohort included 1011 patients without retinopathy and 70 with confirmed retinopathy.

Methods: Participants underwent swept-source OCT, fundus autofluorescence, and Humphrey VF testing using the 10-2 Swedish Interactive Thresholding Algorithm (SITA) Fast, 24-2C SITA Faster, and/or 30-2 SITA Fast protocols. Retinopathy was defined according to the American Academy of Ophthalmology guidelines as the presence of at least 2 abnormalities on structural and/or functional tests. Sensitivity and specificity of each VF protocol were evaluated against OCT as the gold standard.

Main outcome measures: Sensitivity and specificity of the 10-2, 24-2C, and 30-2 protocols.

Results: In early cases without definitive OCT changes (ellipsoid zone loss), parafoveal retinopathy was detected in 71.4% with the 10-2, 57.1% with the 24-2C, and 28.6% with the 30-2. Early pericentral defects were detected on the 24-2C and 30-2 in 57.1% versus 9.5% on the 10-2. The 24-2C’s sensitivity matched that of the 10-2 for parafoveal disease (94.7%) and was comparable with the 30-2 for pericentral disease (90.8% vs 85.5%). Receiver operating characteristic analysis using the 24-2C yielded areas under the curve of 0.844 for mean deviation, 0.881 for pattern standard deviation, and 0.887 for VF index in detecting retinopathy. The mean test duration for 24-2C was 164.8 ± 40.2 seconds, significantly shorter than for 30-2 (253.2 ± 79.1 seconds; P < 0.001) among eyes that underwent both tests; among those tested with both 10-2 and 24-2C, 24-2C was also faster than 10-2 (237.4 ± 64.8 seconds; P = 0.001).

Conclusions: The Humphrey 24-2C protocol offers a reliable, time-efficient, and single-test solution for screening both parafoveal and pericentral HCQ retinopathy, with diagnostic performance comparable with that of the 10-2 protocol for parafoveal retinopathy and 30-2 protocol for pericentral one.