Public Health.

Background: Survival estimates for individuals with Alzheimer's disease (AD) are informative to understand the full disease trajectory. A previous meta-analysis estimated the mean survival of AD patients at 5.8 years from diagnosis, but precise estimates for atypical AD variants are scarce. Atypical AD variants are characterized by a non-amnestic phenotype, a relatively early-onset and lower prevalence of APOE4, which are all possible modulators of clinical trajectories. Here we aimed to evaluate how phenotypical atypicality affects mortality for posterior cortical atrophy (PCA; visual-AD), logopenic variant primary progressive aphasia (lvPPA; language-AD) and behavioral-AD (bvAD) versus a typical AD reference group while taking other risk-modulators into account.

Method: 2084 amyloid-positive cases from the Amsterdam Dementia Cohort were grouped into typical AD (n = 1802) versus atypical AD (n = 282; PCA [n = 112], lvPPA [n = 88], and bvAD [n = 82]) cases. Kaplan-Meier analysis was performed to obtain survival estimates from diagnosis to death. Cox proportional-hazard models were performed to assess effects of atypical AD on mortality, using typical AD as the reference group. We modeled atypical AD subtypes separately (model-1) or combined (model-2) while accounting for the potential effects of age, sex, education, MMSE and APOE4 on mortality. A Likelihood-ratio test was performed to assess whether model-2 with atypical AD provided a better model-fit than model-2 without atypical AD.

Result: Table-1 provides sample characteristics. Median survival times were 6.3[95%CI:5.7-7.6] for lvPPA and 6.0[5.4-6.9] for PCA, compared to 7.0[6.7-7.2] for typical AD. The median survival time for bvAD 5.9[4.3-9.0] could only be extrapolated given that 44% had deceased (Figure-1a). Adjusted median survival times were 7.0[6.7-7.2] for typical and 6.1[5.7-6.5] for atypical AD as a combined group (Figure-1b). In model-1, PCA (HR = 1.37[1.06-1.78]), male sex (HR = 1.44[1.28-1.62]), age (HR = 1.02[1.01-1.03]), and lower MMSE (HR = 0.93[0.92-0.94]) were associated with increased mortality risk (Table-2). All atypical AD variants combined were associated with increased mortality risk (HR = 1.36[1.14-1.63]) compared to typical AD, and addition of atypical AD to the model significantly improved the model-fit (X = 10.5;p<0.01).

Conclusion: Survival in atypical AD is shorter compared to typical AD, even when correcting for other risk-modulators. Hence, atypicality is an important risk factor for mortality beyond age, sex, education and MMSE.