Mixed Neuropathology and Associations with Cognitive Impairment in Autopsied Older Adults

Abstract

Background: Dementia represents a heterogeneous group of clinical syndromes most often driven by multiple coexisting neuropathologic processes. Although Alzheimer’s disease neuropathologic change (ADNC) remains the most recognized contributor, other neurodegenerative and vascular pathologies, including limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC), Lewy body disease (LBD), hippocampal sclerosis, and cerebrovascular lesions, are common and substantially influence cognitive outcomes. Understanding how these mixed pathologies cluster together and contribute to resilience or resistance to cognitive impairment is essential for refining diagnostic frameworks and developing targeted interventions.This dissertation aimed to characterize the heterogeneity of neuropathologic disease and its relationship to clinical outcomes through three complementary aims: (1) to identify biologically coherent clusters of mixed neuropathologic profiles using unsupervised clustering methods; (2) to determine how non-Alzheimer’s pathologies contribute to resistance and resilience to ADNC; and (3) to investigate neuropathologic correlates of resilience to LATE-NC across multiple autopsy cohorts. Methods: Chapter 2 applies unsupervised hierarchical clustering to 2,899 autopsied National Alzheimer’s Coordinating Center (NACC) participants using fourteen neuropathologic features. Chapter 3 examines non-AD pathologies among NACC participants classified as resistant (no/low ADNC, cognitively normal), resilient (intermediate/high ADNC, cognitively normal), or impaired (intermediate/high ADNC, cognitively impaired), using multivariable logistic regression and longitudinal cognitive assessments. Chapter 4 examines factors associated with resilience to advanced LATE-NC (stage 2–3) in both NACC and Adult Changes in Thought (ACT) study participants aged ≥75 years, using regression models stratified by age and longitudinal cognitive analyses in ACT. Results: Across Chapters 2–4, mixed neuropathologic disease strongly influenced cognitive outcomes. In Chapter 2, five distinct clusters of neuropathologic disease were identified, reflecting common combinations of ADNC, LATE-NC, LBD, TDP-43, and vascular pathology. These clusters were biologically interpretable and often aligned with traditional clinical diagnoses. In Chapter 3, both resistance and resilience to ADNC were associated with markedly lower burden of non-AD pathologies, particularly LBD, LATE-NC, hippocampal sclerosis, and arteriosclerosis. In Chapter 4, resilience to LATE-NC was linked to reduced ADNC and absence of hippocampal sclerosis. Across analyses, lower mixed pathologic burden was strongly associated with cognitive resilience. Conclusions: Across studies, cognitive outcomes depended on the cumulative and interactive effects of multiple neuropathologic processes. These findings highlight the limitations of single-pathology diagnostic frameworks and support development of biologically grounded subtyping that accounts for mixed pathology. Such approaches will be critical for improving diagnostic precision, understanding mechanisms of resilience, and guiding biomarker and therapeutic development.