Performance of the Interleukin Score for IntraOcular Lymphoma Diagnosis Score in the Diagnosis of Vitreoretinal Lymphoma in Clinical Practice.

Abstract

To evaluate the real-world diagnostic performance of the Interleukin Score for IntraOcular Lymphoma Diagnosis (ISOLD) score-a probabilistic formula based on absolute intraocular interleukin-10 (IL-10) and IL-6 levels-and to compare it with absolute cytokine thresholds and the IL-10/IL-6 ratio for vitreoretinal lymphoma (VRL) diagnosis. Because the comparison group consists predominantly of patients who underwent anterior chamber paracentesis (ACP), this study primarily assesses ISOLD performance on aqueous humor (AH) samples while providing additional data on vitreous specimens. Retrospective comparative diagnostic analysis. A total of 166 patients who underwent intraocular sampling for IL-6 and IL-10 measurement at a tertiary VRL referral center between January 2017 and June 2019, yielding 183 samples (156 AH from ACP and 27 vitreous). Furthermore, 25 patients had confirmed VRL (18 AH samples, 20 vitreous samples; 13 had both, 7 vitreous only, and 5 AH only). The comparison group comprised 141 non-VRL patients: 9 patients with confirmed CNS lymphoma without ocular involvement (PCNSL+ non-VRL group, 9 AH only) and 132 patients without lymphomatous disease (non-PCNSL group, 129 AH and 7 vitreous). IL-6 and IL-10 were measured in undiluted intraocular samples by cytometric bead array. The ISOLD score was calculated using published formulas specific to AH and vitreous. Diagnostic performance (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV]) was assessed for the ISOLD score and the IL-10/IL-6 ratio, stratified by sample type. Receiver operating characteristic (ROC) curve analysis was performed to identify optimal thresholds. Sensitivity, specificity, PPV, and NPV of the ISOLD score and the IL-10/IL-6 ratio for VRL diagnosis were reported separately for AH and vitreous samples. At a threshold of 0, the ISOLD score yielded an overall sensitivity of 86.8%, specificity of 98.6%, PPV of 94.3%, and NPV of 96.6% across all 183 samples. For AH samples (n = 156), which constitute most of this data set, sensitivity, specificity, PPV, and NPV were 77.8%, 98.6%, 87.5%, and 97.1%, respectively; the optimal AH threshold by ROC analysis was -5.2, yielding a sensitivity of 94.4%, specificity of 96.7%, PPV of 73.9%, NPV of 99.3%, and an area under the curve (AUC) of 0.966. For vitreous samples (n = 27), performance at a threshold of 0 was 95.0%, 100%, 100%, and 87.5%, respectively; the optimal vitreous threshold was -2.55, yielding 100% sensitivity, specificity, PPV, and NPV, with an AUC of 1.0. Two false positives occurred in varicella zoster virus-associated ocular disease; 5 false negatives were observed, predominantly in patients receiving systemic corticosteroids and/or chemotherapy. The IL-10/IL-6 ratio at a threshold of 0.6 in AH (n = 150) demonstrated a sensitivity of 77.8%, specificity of 93.9%, PPV of 63.6%, and NPV of 96.9%; at a threshold of 1 in vitreous (n = 27), sensitivity, specificity, PPV, and NPV were 80%, 100%, 100%, and 63.6%, respectively. The optimal AH threshold by ROC analysis for the IL-10/IL-6 ratio was 0.48, yielding a sensitivity of 83.3%, specificity of 93.9%, PPV of 65.2%, NPV of 97.6%, and an AUC of 0.942. For vitreous samples, performance at a threshold of 0.195 was 100%, 100%, 100%, and 100%, respectively, with an AUC of 1.0. Bilateral involvement and intermediate uveitis were significantly more frequent in the VRL group (P = .008 and P = .005, respectively). The ISOLD score is a reliable tool for VRL diagnosis in routine clinical practice, with performance validated primarily on AH samples obtained by ACP. It outperforms the IL-10/IL-6 ratio alone by providing a continuous probabilistic estimate. False-negative results may occur in immunosuppressed patients or those receiving corticosteroids or chemotherapy. A larger vitreous sample would be required to draw definitive conclusions regarding ISOLD performance on this specimen type. Cytological analysis of vitreous samples remains the gold standard for definitive VRL confirmation.