LAUSR

We read the comment by Uleman and colleagues on our article1 with interest. The details on the AD criteria of the individual studies can be seen in supplementary table e-6 of our publication. Briefly, most neuroimaging studies diagnosed their AD patients with the NINCDS-ADRDA and DSM-IV criteria. Only Risacher et al.2 performed biomarker-based diagnosis. Postmortem studies relied on neuropathologic diagnosis of AD. We thus believe that this issue may only make a minimal contribution to the heterogeneity found. Instead, our recent study shows that a major source of heterogeneity is variability across studies in subtyping methods.3 We agree that the link between resilience/resistance and risk factors and comorbidity is through complex pathways. In our study, we did not elaborate on whether the severity and typicality axes are linear or nonlinear. Indeed, we believe that all—if not most—of the factors included in our model relate with each other in intricate associations, not only resilience and resistance. The next step is to empirically test and expand this conceptual model. Methods such as the ones mentioned by Uleman and colleagues will be important to understand associations among multiple factors in our conceptual model, helping to unravel the heterogeneity within AD.