We interpret this as follows: the discovery cohort had an extremely small sample size (only 6 pooled samples), and FDR correction tends to be overly conservative; however, all three candidate proteins exhibited highly significant differences ( P < 0.001) in ELISA assays conducted in an independent large-sample validation cohort ( n = 260).
0.845, P = 0.0013), and the point estimates of NRI and IDI indicated improved clinical significance; (2) Multicollinearity tests revealed that the VIF values for all variables were <1.2, indicating a stable and reliable model; (3) Although FDR-corrected results did not reach statistical significance due to the small sample size of the discovery cohort, independent large-scale ELISA validation provided decisive evidence for the reliability of the candidate proteins.