Disease duration in autosomal dominant familial Alzheimer disease: A survival analysis

Abstract

Objective: To use survival modeling to estimate disease duration in autosomal dominant familial Alzheimer disease (ADAD) and ascertain whether factors influencing age at onset also affect survival.

Methods: Symptomatic mutation carriers (201 presenilin 1 [PSEN1] and 55 amyloid precursor protein [APP]) from ADAD families referred to the Dementia Research Centre, between 1987 and 2019, were included. Survival was assessed with respect to age at onset, year of birth, APOE ε4 status, cognitive presentation, and sex using multilevel mixed-effects Weibull survival models. The contribution of mutation and family to variance in age at onset and duration was also assessed.

Results: Estimated mean survival was 11.6 (10.4-12.9) years and was similar for APP and PSEN1 mutations. Sixty-seven percent of the variance in age at onset was explained by mutation and 72% by mutation and family together. In contrast, only 6% of the variance in disease duration was explained by mutation specificity and 18% by family membership. Irrespective of gene, survival appeared longer for successive generations and in individuals with atypical presentations. Older age at onset was associated with longer duration within PSEN1 and shorter duration within APP mutation carriers. No differences in survival time were found between sexes or between mutations located before or beyond codon 200 within PSEN1.

Conclusions: Survival is influenced by mutation to a much lesser extent than age at onset. Survival time has increased over time and is longer in atypical presentations. These insights may inform the interpretation of disease-modifying therapy trials in ADAD.

Figure 1. Symptom onset, age at death, disease duration, and survival probability by gene

(A) Unadjusted Kaplan-Meier survival plots show the estimated survival probability by disease duration for PSEN1 vs APP. The blue line references APP and the red line PSEN1. Ninety-five percent CIs and number of individuals still alive per disease duration length, by 10 years, by 20 years, and by 30 years, are also shown. (B) Violin plots show the distribution of age at symptom onset, at death, and disease duration for PSEN1 vs APP. Data are median (line) with median interquartile range (upper and lower dotted lines). Age at onset: 42 (38–48) years vs 50 (48–55) years; age at death: 52 (46–58) years vs 61 (58–66) years; and disease duration: 8 (6–12) years vs 10 (8–13) years. “*” indicates significant difference between groups. (C) Scatter plot shows the association between age at symptom onset and age at death in PSEN1 vs APP. The solid line represents the line of best fit from the survival model, adjusted for sex, year of birth, and clustered by family membership for each gene. The shaded area represents 95% CIs. Markers show the unadjusted raw data: hollow blue triangles represent individuals with APP mutations and hollow red circle markers individuals with PSEN1 mutations. APP = amyloid precursor protein; CI = confidence interval; PSEN1 = presenilin 1.

Figure 2. Survival probability pre- and post-births in the 1930s

Unadjusted Kaplan-Meier survival plot showing survival by disease duration for individuals born before and after the 1930s. Green references individuals born by 1930 and orange after 1930. Ninety-five percent CIs and number of individuals still alive per disease duration length, by 10 years, by 20 years, and by 30 years, are also shown. CI = confidence interval; DOB = date of birth.

Figure 3. PSEN1: survival probability by clinical presentation

Unadjusted Kaplan-Meier survival plot shows the estimated survival probability by disease duration for clinical presentations. Ninety-fivepercent CIs and number of individuals still alive per disease duration length, by 10 years, by 20 years, and by 30 years, are also shown. Green references individuals with amnestic presentations and orange individuals with atypical presentations. CI = confidence interval; PSEN1 = presenilin 1.