A monocentric IMAP/IMAP+ sample of 94 cognitively healthy adults (mean age 57.41, SD 18.31; 55% female) contributed up to three neuroimaging visits across a three-year window. The study team collected self-report measures of repetitive negative thinking (RNT) and also recorded anxiety and depressive symptoms using validated instruments; the abstract did not report the specific instrument names or scoring thresholds. Neuroimaging outcomes were predefined as longitudinal change in global cortical amyloid using Florbetapir-PET, cerebral glucose metabolism using FDG-PET, and grey matter volume via MRI. PET images underwent partial volume correction and analyses used masks targeting AD-relevant regions, according to the authors.
The investigators reported that higher baseline RNT scores were associated with a steeper increase in global cortical amyloid over three years. The primary model produced an estimated effect size of B = 0.0120 with a 95% confidence interval from 0.003 to 0.021 and a P value of 0.012. That association remained when analyses were restricted to older subgroups — the authors performed sensitivity checks limited to participants aged ≥50 and ≥60, and data indicated the effect persisted in those subsets. The abstract did not list every covariate included in the model; it did state analyses adjusted for potential confounders, but did not enumerate them in the abstract text.
Voxelwise, FDR-corrected analyses identified regional patterns underlying the global effect. Data indicated that higher RNT predicted significantly faster amyloid accumulation in brain areas commonly implicated early in Alzheimer disease pathology, specifically the precuneus, posterior cingulate cortex, and medial frontal cortex. The study team used AD-relevant neuroanatomical masks for these voxelwise tests; the abstract reported these region-specific signals without providing full voxelwise statistic maps in the abstract itself.
RNT did not associate with longitudinal change in FDG-PET measures of glucose metabolism or with MRI measures of grey matter atrophy (reported Ps > 0.05). Parallel models testing anxiety and depressive symptom scores also returned null findings for change in any neuroimaging outcome (Ps > 0.05), according to the authors. Thus, within the three-year follow-up window and using these imaging modalities, the signal related to RNT was selective to amyloid accumulation.
To aid visualization, the investigators divided RNT scores into tertiles in post-hoc tests. Those comparisons revealed a statistically significant difference in rate of amyloid accumulation between the lowest and highest RNT tertiles (P = 0.008). The primary linear effect (B = 0.0120) and the tertile contrast point to a graded association across the cohort. The authors applied FDR correction for voxelwise analyses to control for multiple comparisons; the abstract reports FDR-corrected regional results but does not reproduce full corrected statistic tables or cluster sizes in the abstract.
This report is posted as a medRxiv preprint and has not been peer reviewed. The abstract discloses limited methodological detail: it does not specify the exact RNT instrument used, the full list of covariates, model specification details, or the quantitative thresholds used for amyloid positivity. The cohort was monocentric and community‑dwelling; investigators noted ethical approval by the CPP Nord-Ouest III ethics committee and stated that data are available on reasonable request pending coordinator approval. The authors argued that because RNT is responsive to psychological treatment, future work should test whether modifying RNT alters amyloid trajectories or clinical risk — however, the presented data are observational and cannot establish causality. These are the concrete unresolved questions the authors flagged: whether RNT is a marker or driver of amyloid pathology, whether intervention can change imaging trajectories, and how covariates and longer follow-up might influence the observed relationships.