Combining biomarkers to predict the disease activity of graves' ophthalmopathy: a combinatory model of the NLR, TRAb and FT4.

PURPOSE: The aim of this study was to identify potential biomarkers associated with active Graves' ophthalmopathy (GO) and develop a model for predicting the occurrence and progression of active GO.

METHODS

A retrospective study was conducted on 220 GO patients (n=120 in the active phase, n=100 in the stable phase) and 70 healthy controls. Laboratory and other clinical indicators were compared in GO patients at different stages and healthy controls. A multivariate regression analysis model was used to analyze the clinical risk factors affecting the occurrence and progression of active GO, and a predictive model based on risk factors was established.

RESULTS

Higher WBC, NEU, NLR, TRAb and FT4 levels (all p<0.005) were detected in patients with active GO than those in inactive GO. The CAS score of the GO patients was positively correlated with the WBC (r = 0.155, p = 0.003), NEU (r = 0.165, p = 0.002), NLR (r = 0.134, p = 0.010), FT3 (r = 0.117, p = 0.031), FT4 (r = 0.139, p = 0.011), and TRAb (r = 0.160, p = 0.004) counts and negatively correlated with the TSH level (r = - 0.114, p = 0.043). Multivariate regression analysis revealed that the NLR, TRAb, and FT4 were significant risk factors for GO progression (all p < 0.05). The cut-off levels for predicting active GO were 8.71 IU/L for TRAb (AUC=0.643, sensitivity 0.58, specificity 0.74), 19.82 ng/dL for FT4 (AUC=0.606, sensitivity 0.41, specificity 0.75), and 2.405 for the NLR (AUC=0.597, sensitivity 0.50, specificity 0.75). A predictive model including these risk factors was built (the cutoff value was 0.6, the AUC was 0.716, the sensitivity was 0.533, and the specificity was 0.808).

CONCLUSION

The laboratory biomarkers for the occurrence and progression of active GO include the NLR, FT4, and TRAb. We have developed a predictive combined model that may assist in timely assessment of GO activity and progression, and provide clues for future development of facile tools to predict GO activity.