Table 4. Continuous net reclassification improvement (NRI) of risk stratification of fatty liver after adding hsa-miR-122-5p and hsa-885-5p to the different base models* including conventional risk factors.
| Model/microRNAs | AUC | NRI (95% CI) | p-value |
|---|---|---|---|
| Model 1 (predicting FLCvs. NL) | 0.983 | Reference | Reference |
| +hsa-miR-122-5p | 0.986 | 0.963 (0.570–1.356) | 1.6 * 10−6 |
| +hsa-miR-885-5p | 0.983 | 0.139 (−0.296–0.575) | 0.53 |
| +hsa-miR-122-5p and hsa-miR-885-5p | 0.987 | 1.363 (1.057–1.670) | 2.8 * 10−18 |
| Model 2 (predicting FLAll vs. NL) | 0.897 | Reference | Reference |
| +hsa-miR-122-5p | 0.898 | 0.125 (−0.080–0.329) | 0.23 |
| +hsa-miR-885-5p | 0.897 | 0.055 (−0.1515–0.262) | 1.00 |
| +hsa-miR-122-5p and hsa-miR-885-5p | 0.898 | 0.278 (0.077–0.480) | 0.0067 |
*Statistical model: Stepwise models 1 and 2 consist of all variables associated with FL in the Young Finns Study [see ref. 6].
Variables with independent and significant prediction value were: Model 1: Alanine aminotransferase (ALT), gamma-glutamyltransferase (GT), waist circumference, insulin levels predicting individuals with FLC vs. NL (NL n = 524, FLC n = 21).
Model 2: ALT, GT, waist circumference, insulin levels, and systolic blood pressure predicting individuals with FLAll vs NL (NL n = 524, FLAll n = 108).
Abbreviations: AUC = area under curve, NL = normal liver, FL = fatty liver, FLAll = all subjects with FL, FLC = clearly identified fatty liver.