It is an open issue whether blood biomarkers serve to diagnose Alzheimer's disease (AD) or monitor its progression over time from prodromal stages. Here, we addressed this question starting from data of the European FP7 IMI-PharmaCog/E-ADNI longitudinal study in amnesic mild cognitive impairment (aMCI) patients including biological, clinical, neuropsychological (e.g., ADAS-Cog13), neuroimaging, and electroencephalographic measures. PharmaCog/E-ADNI patients were classified as "positive" (i.e., "prodromal AD" n = 76) or "negative" (n = 52) based on a diagnostic cut-off of Abeta42/P-tau in cerebrospinal fluid as well as APOE epsilon 4 genotype. Blood was sampled at baseline and at two follow-ups (12 and 18 months), when plasma amyloid peptide 42 and 40 (Abeta42, Abeta40) and apolipoprotein J (clusterin, CLU) were assessed. Linear Mixed Models found no significant differences in plasma molecules between the "positive" (i.e., prodromal AD) and "negative" groups at baseline. In contrast, plasma Abeta42 showed a greater reduction over time in the prodromal AD than the "negative" aMCI group (p = 0.048), while CLU and Abeta40 increased, but similarly in the two groups. Furthermore, plasma Abeta42 correlated with the ADAS-Cog13 score both in aMCI patients as a whole and the prodromal AD group alone. Finally, CLU correlated with the ADAS-Cog13 only in the whole aMCI group, and no association with ADAS-Cog13 was found for Abeta40. In conclusion, plasma Abeta42 showed disease progression-related features in aMCI patients with prodromal AD.