Abstract
Purpose: To determine if tear biomarker analysis could be utilized as an objective diagnostic test for dry eye disease (DED).
Methods: Normal (n=18) and DED (n=60) subjects were recruited. DED status was established using a validated symptom questionnaire (OSDI), non-invasive tear break-up time and Schirmer test. A tear sample (1µl) was collected from each subject and analyzed for 7 inflammatory markers using a multiplex immunoassay. Receiver operator characteristic (ROC) curves were used to establish sensitivity and specificity values and the most appropriate diagnostic cut-off for DED for each biomarker. ROC curves were also used for established clinical tests (tear osmolarity, evaporation rate, tear turnover rate (TTR) and assessment of corneal staining) for comparison to tear biomarker analysis.
Results: The results show that, in terms of sensitivity and specificity(and when considered individually) the 7 biomarkers are not as accurate at diagnosing symptomatic DED, as the established clinical tests. The most promising biomarker, tumour necrosis factor alpha, showed sensitivity of 61% and specificity of 61%, at a point of 2930 pg/ml. The other 6 biomarkers showed lower predictive power. However when considered together as a panel of 7, the biomarkers sensitivity was 70% and specificity 67%. This indicated a better predictive power for DED than clinical tests, e.g. tear evaporation rate (56% sensitivity, 50% specificity) and tear osmolarity (62% sensitivity, 56% specificity), and a similar result to TTR (73% sensitivity, 67% specificity). Assessment of corneal staining was the best predictor for DED, with sensitivity of 85% and specificity of 94%. However, corneal staining is a subjective test and vulnerable to intra/inter-examiner variability.
Conclusions: A panel of tear biomarkers have proven to be comparable to established clinical tests for predictive power of DED. However, no individual biomarker indicative of DED has been determined. Thus, further research investigating other potential
biomarkers is required.
Methods: Normal (n=18) and DED (n=60) subjects were recruited. DED status was established using a validated symptom questionnaire (OSDI), non-invasive tear break-up time and Schirmer test. A tear sample (1µl) was collected from each subject and analyzed for 7 inflammatory markers using a multiplex immunoassay. Receiver operator characteristic (ROC) curves were used to establish sensitivity and specificity values and the most appropriate diagnostic cut-off for DED for each biomarker. ROC curves were also used for established clinical tests (tear osmolarity, evaporation rate, tear turnover rate (TTR) and assessment of corneal staining) for comparison to tear biomarker analysis.
Results: The results show that, in terms of sensitivity and specificity(and when considered individually) the 7 biomarkers are not as accurate at diagnosing symptomatic DED, as the established clinical tests. The most promising biomarker, tumour necrosis factor alpha, showed sensitivity of 61% and specificity of 61%, at a point of 2930 pg/ml. The other 6 biomarkers showed lower predictive power. However when considered together as a panel of 7, the biomarkers sensitivity was 70% and specificity 67%. This indicated a better predictive power for DED than clinical tests, e.g. tear evaporation rate (56% sensitivity, 50% specificity) and tear osmolarity (62% sensitivity, 56% specificity), and a similar result to TTR (73% sensitivity, 67% specificity). Assessment of corneal staining was the best predictor for DED, with sensitivity of 85% and specificity of 94%. However, corneal staining is a subjective test and vulnerable to intra/inter-examiner variability.
Conclusions: A panel of tear biomarkers have proven to be comparable to established clinical tests for predictive power of DED. However, no individual biomarker indicative of DED has been determined. Thus, further research investigating other potential
biomarkers is required.
Original language | English |
---|---|
Number of pages | 1 |
Publication status | Published - 20 Sept 2016 |