Reply to “Clarifying the Moderating Role of Socioeconomic Status on Amyloid‐Related Gray Matter Atrophy”

To the Editor,

We thank Drs. Zhou, Peng, and Wang for their interest and thoughtful comments on our recent article, “Protective Effects of Socioeconomic Status and Lifestyle on Amyloid‐ and White Matter Hyperintensity‐Related Longitudinal Brain Atrophy and Cognitive Decline.” ¹ We appreciate the opportunity to clarify and expand on several points raised in their letter.

First, we acknowledge the inconsistency between the caption of figure 4 and the main text. The caption incorrectly states that “The mediating effect of gray matter (GM) atrophy is more pronounced in individuals with high socioeconomic status (SES)…” However, as indicated in figure 4 and correctly described in the text, the association between amyloid (Aβ) positron emission tomography (PET) burden and Aβ‐related GM atrophy is more pronounced in individuals with low SES. Importantly, all other references to this effect are accurate, and this error does not alter our interpretation or conclusions. We apologize for the oversight, and we thank the authors of the letter for their careful reading of our study. A correction is published with this reply.

Second, in our original analysis we applied a median split for the moderated mediation analysis to simplify both the analysis and visualization of our results. We consider it unlikely that this approach biased our findings, as the initial models predicting cognitive performance treated SES as a continuous variable and similar patterns also emerged when we repeated the analysis separately in men and women using gender‐specific medians (see Fig S4 in the Supplementary Materials in our original article). To further clarify, we now conducted a Johnson–Neyman analysis (Fig 1A), which produced results consistent with our initial median split approach, while providing additional information by identifying that the interaction between Aβ standardized uptake value ratio and education was no longer significant above 18 years of education. The low‐ and high‐SES group‐specific coefficients and confidence intervals are already provided in the mediation models (Figures 4C and 4E in our original article).

FIGURE 1.

Interactive effects between amyloid β (Aβ) standardized uptake value ratio (SUVR) and socioeconomic status (SES) indicators on the rate of Aβ‐related gray matter atrophy. (A) Johnson–Neyman plot showing that the interaction between Aβ SUVR and years of education does not significantly predict gray matter atrophy rates when education exceeds 18 years. (B–D) Interactions between Aβ SUVR and individual SES indicators on atrophy rates, estimated in separate linear regression models adjusted for age, sex, and APOE4 status. SES indicators were standardized for analysis. For the model including net income, 22 of 181 participants had to be excluded because of missing data. Analyses were conducted using the interactions package in R. [Color figure can be viewed at www.annalsofneurology.org]

Third, in our study we modeled SES as a latent variable to capture its multi‐dimensional nature and reduce multiple comparisons. We agree that examining individual SES indicators is informative, particularly because years of education had a very high factor loading. In response to the letter, we, therefore, ran separate linear regression models (adjusted for age, sex, and APOE4 status) testing interactions between Aβ PET burden and years of education, occupation, and net income. Educational complexity was not included separately because of its strong correlation with years of education (Spearman's ρ = 0.81, Fig 1C in our original article). These analyses (Fig 1B–D) suggest that both years of education and net income, but not occupation, moderate Aβ‐related GM atrophy. When including both interactions in the same model, Aβ × net income remained significant (β = 0.0026, p = 0.016), while Aβ × years of education was marginal (β = 0.002, p = 0.052).

Together, these complementary analyses support the robustness of our findings and provide tentative evidence that education and net income independently influence the rate of Aβ‐related GM atrophy. Larger studies will be important to confirm these results.

Author Contributions

Dario Bachmann: Visualization; writing – original draft; formal analysis. Valerie Treyer: Writing – review and editing.