Noncoding RNAs as novel diagnostic biomarkers for diabetic retinopathy: A systematic review and meta-analysis.
Ting Zhang, Donghong Xu
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis assessed the diagnostic accuracy of noncoding RNAs (ncRNAs) for diabetic retinopathy (DR), pooling 24 studies from 17 articles involving 1,983 patients and 1,437 controls. Overall, ncRNAs had a pooled sensitivity of 0.81, specificity of 0.84, and area under the curve of 0.89. Long noncoding RNAs (lncRNAs) showed better diagnostic performance than microRNAs (P = 0.03). The authors report substantial heterogeneity, while sensitivity analyses supported the robustness of the pooled findings and no notable publication bias was detected.
Key findings
- Pooled sensitivity was 0.81 and specificity was 0.84 for ncRNAs in diagnosing diabetic retinopathy.
- The pooled area under the curve was 0.89; positive and negative likelihood ratios were 4.96 and 0.23, respectively.
- LncRNAs had superior diagnostic performance compared with miRNAs (P = 0.03).
- Significant heterogeneity was observed (I 2 > 50%), although sensitivity analysis supported robustness.
Why it matters
Early detection is important in diabetic retinopathy, and these pooled results suggest circulating ncRNAs may help distinguish people with DR from controls. The reported diagnostic performance, particularly for lncRNAs, supports further evaluation of these markers as noninvasive screening or diagnostic tools. However, pooled accuracy does not by itself establish how well they perform in routine clinical settings.
What to keep in mind
Significant heterogeneity was present across the included studies (I 2 > 50%), which may limit how consistently the pooled diagnostic estimates apply across settings.