To evaluate the association between longitudinal GM changes and basal CSF Alzheimer's Disease biomarker levels on elderly cognitively normal subjects, AD preclinical subjects (Aβ42 positive) and healthy controls. Two independent cohorts have been analyzed to improve the generalizability of results.The sample consisted in 62 cognitively healthy subjects (MMSE > 28, CDR = 0 and objective cognitive performance within the normal range), aged 60-80. Thirty-one of them were recruited at Hospital Clinic of Barcelona and the rest were selected from ADNI dataset (http://adni.loni.usc.edu). All subjects had been submitted to two MRI sessions delayed 2-years apart and to a lumbar punction at the time of the first scan to obtain CSF Aβ42 and p-tau levels. Eighteen subjects resulted Aβ42 positive (preclinical AD) and 26 had abnormally high value of p-tau, based on previously reported normality thresholds for the respective cohorts. The MRI protocol for the two sessions was in all cases exactly the same and contained a high resolution 3d T1-weigthed imaging (1x1x1mm3 in local sample and 1x1x1.2mm3 in subjects from ADNI dataset). Demographics are shown in table 1. Pre and post images were corregistered with longitudinal pair-wise registration tool from SPM12. Corregistration resulted in a pre-post average image and an image with the divergence of the pre-post matching velocities, which accounted for local longitudinal changes of volume. Average images were normalized to MNI through DARTEL and resulting warps were applied to GM-WM masked divergence images, that later were smoothed with an 8 mm FWHM Gaussian kernel. Images were then divided by the exact time between scans (atrophy rates) and masked with a common GM mask. Voxel-wise correlations of atrophy rates with Aβ42, p-tau and p-tau/Aβ42 ratio, after correction by age, gender and educational level, were carried out. Statistical significance was considered for p<0.001 uncorrected for multiple comparisons and a minimum cluster size of 100 voxels.Figure 1 shows the maps of the regions that reached statistically significant associations with the core AD CSF biomarkers. Aβ42 and p-tau maps were disjoint. Atrophy rates were correlated with Aβ42 in medial and orbital frontal, precuneus, cingulate, medial temporal regions and cerebellum, whereas, correlations with p-tau were located in left hippocampus, parahippocampus and striatal nuclei. On the other hand, p-tau/Aβ42 ratio maps shared areas with Aβ42 and p-tau maps, but also showed specific regions mainly in temporal medial structures and left basal ganglia. All the regions found in the correlation analysis have been widely reported to be associated to AD. In the preclinical stage of AD, different pathological mechanisms might underpin atrophy in diverse brain areas known to be related to AD. At this stage of the disease, hippocampal atrophy seems to be independent on Aβ42 and just dependent on p-tau levels.